Safety and efficacy of the non-cessation method of antithrombotic agents after emergency endoscopic hemostasis in patients with non-variceal upper gastrointestinal bleeding: A multicenter pilot study

Background and Aims The present study aimed to clarify the safety and efficacy of the non-cessation method of antithrombotic agents after emergency endoscopic hemostasis in Japanese patients with non-variceal upper gastrointestinal bleeding (UGIB). Methods In this multicenter, prospective, pilot study, we performed emergency endoscopic hemostasis for non-variceal UGIB in patients taking antithrombotic agents and resumed the medications without a cessation period (group A). The clinical characteristics, types of antithrombotic agents, UGIB etiology, treatment outcome, and adverse events were evaluated. We used propensity score matching to compare treatment outcome and adverse events with our previous cohort (group B) in whom antithrombotic agents were transiently discontinued after emergency endoscopic hemostasis. Antithrombotic agents possibly can be continued after successful emergency endoscopic hemostasis for non-variceal UGIB.


Introduction
Non-variceal upper gastrointestinal bleeding (UGIB) is one of the most common causes of gastrointestinal emergencies, sometimes requiring emergency endoscopic hemostasis. While Helicobacter pylori (H. pylori) infection and nonsteroidal anti-inflammatory drugs are the two main causes of hemorrhagic peptic ulcer disease [1][2][3][4][5], an alteration in the epidemiology and pathophysiology of UGIB has been recently suggested because of the decreasing prevalence of H. pylori infection as well as the increasing use of antithrombotic agents [6,7].
Although it remains uncertain whether antithrombotic agents (other than aspirin) can cause gastrointestinal mucosal damage, those medications have been reported to increase the risk of UGIB [8,9], and we determined the actual incidence of non-variceal UGIB to be around 1.0% in Japanese patients taking antithrombotic agents [10]. Furthermore, patients taking antithrombotic agents exhibit more severe clinical signs of non-variceal UGIB [11]. It therefore seems important to determine the optimal clinical management of non-variceal UGIB, especially in patients requiring emergency endoscopic hemostasis.
Theoretically, antithrombotic agents should be stopped after endoscopic hemostasis considering the risk of recurrent UGIB. However, cessation of antithrombotic agents exposes the patients to the risk of thromboembolic events because of their underlying condition [12]. Such thromboembolic events can occur even when antithrombotic agents were discontinued for only 5-7 d [13]. Considering these conflicting risks, the Japan Gastroenterological Endoscopy Society (JGES) released new guidelines in 2012 for the management of patients on antithrombotic agents with regard to endoscopic procedures [14]. This revised JGES guideline focused more on avoiding thromboembolic events than GI bleeding adverse events.
Briefly, according to the previous JGES guideline published in 2005 [15], cessation of antiplatelet agents was generally recommended for a certain period before invasive gastrointestinal endoscopy (e.g., 5 d for antiplatelet monotherapy and 7 d for combined antiplatelet therapy). Cessation of warfarin was recommended 3-4 d before any procedures with or without heparin replacement to decrease the international normalized ratio <1.5. In contrast, the 2012 JGES guideline generally recommended that antiplatelet agents be continued before invasive gastrointestinal endoscopy, especially in patients with a higher risk of thromboembolism. In patients on anticoagulant therapy such as warfarin and dabigatran, replacement with heparin was recommended before invasive gastrointestinal endoscopy.
We previously compared the clinical management of patients with non-variceal UGIB before and after the release of the 2012 JGES guideline, and we demonstrated that the cessation period for antithrombotic agents was significantly shortened after the release of the guideline [11]. However, the risk of rebleeding after emergency endoscopic hemostasis without a cessation period has not been clearly demonstrated.
The aims of this pilot study were: (1) to evaluate the safety of the non-cessation method of antithrombotic agents after emergency endoscopic hemostasis in patients with non-variceal UGIB, and (2) to compare the clinical outcome and adverse events with our previous cohort in whom antithrombotic agents were transiently discontinued after emergency endoscopic hemostasis.

Study design and ethical issues
This was designed as a multicenter prospective pilot study. Patients who had undergone emergency endoscopy and endoscopic hemostasis for non-variceal UGIB from February 2018 through October 2019 at participating institutions were enrolled.
Patients >20 years of age who fulfilled the following criteria were candidates for the study: (1) underwent endoscopic hemostasis within 24 h of symptom onset; (2) used antithrombotic agents, including antiplatelet agents and anticoagulants; and (3)

Patient data
Clinical data collected in the prospective cohort were age, gender, alcohol consumption, smoking habit, H. pylori infection, hemoglobin level, comorbidity (including Charlson comorbidity score), and types of medications at study entry (antithrombotic agents, nonsteroidal anti-inflammatory drugs, and antacids). We also accumulated information concerning endoscopic findings, hemostatic procedure, and treatment outcome (transfusion volume, fasting period, and hospitalization period). All patients in the prospective cohort resumed antithrombotic agents after emergency endoscopic hemostasis without a cessation period (group A).
In the present study, we used the clinical data of our retrospective cohort of 154 patients on antithrombotic treatment who underwent emergency endoscopic hemostasis for non-variceal UGIB as group B [11]. These patients were similarly treated but antithrombotic agents were transiently discontinued after endoscopic hemostasis in accordance with the JGES guideline at that time [12,13].

Endoscopic hemostasis
Endoscopic hemostasis was performed mainly by high-frequency soft coagulation or hemoclipping using a forward-viewing water-jet endoscope (GIF-Q260J; Olympus, Tokyo, Japan) [16,17]. The choice of procedure was at the endoscopists' discretion. Soft coagulation was performed using an electrosurgical unit (ICC200, Erbe Elektronedizin GmbH, Tübingen, Germany) and monopolar hemostatic forceps (Radial Jaw TM 4 Hot Biopsy Forceps, Boston Scientific, Tokyo, Japan) inserted through the channel of the endoscope [16]. Briefly, when the bleeding vessel was identified, electric current was gently applied using the closed forceps to the area surrounding the target vessel using the soft coagulation mode (60 W). Once blood flow diminished, the endoscopist coagulated the target vessel using 80 W in gastric ulcers and 60 W in duodenal ulcers [16]. Hemoclipping was performed using hemoclips (HX-610-135, HX-610-135S hemoclips; Olympus Optical, Tokyo, Japan) that were placed on the ulcer base to bind the detected vessels [16,18]. In both procedures, when the ulcer base was covered by massive clots, the clots were initially removed either by water jet or by endoscopic forceps to visualize the vessel(s) responsible. Both procedures were repeated until hemostasis was endoscopically confirmed. Interventional radiology and/or surgery was implemented when the endoscopic hemostasis was considered ineffective.
After endoscopic hemostasis was achieved, all patients were hospitalized and managed conventionally (fasting with peripheral parenteral nutrition and intravenous proton pump inhibitors). The indications for blood transfusion were a hemoglobin level <6 g/dL on admission or a rapid drop in hemoglobin level >2 g/dL in patients with hemoglobin level <10g/dL at baseline. Follow-up endoscopy was performed 48-72 h after the initial endoscopy, and repeat endoscopic hemostasis was applied in case recurrent UGIB was detected. Interventional radiology and/or surgery was implemented when the endoscopic hemostasis was considered ineffective.
All endoscopic hemostasis was performed by 18 endoscopists, comprising eight specialists and 10 trainees. Specialists were defined as endoscopists who had performed endoscopy for more than 5 years with experience in more than 40 endoscopic submucosal dissection (ESD) procedures after mastering the required fundamental skills and knowledge [19,20].

Study outcomes
The primary endpoint of the present study was the rate of rebleeding within a month after emergency endoscopic hemostasis in group A. Rebleeding was defined as follows: (1) follow-up endoscopy identified recurrent UGIB or the stigmata of recent hemorrhage, or (2) melena and progressive anemia with a decrease in hemoglobin level greater than 2g/dL and/or with a decrease in systolic blood pressure <80 mmHg. The secondary endpoint was the incidence of adverse events within a month (rebleeding, thromboembolic events, and mortality) in group A.
To compare the non-inferiority of the non-cessation method of antithrombotic agents to the transient discontinuation method based on the JGES guidelines, treatment outcome and adverse events in group A were compared with those in group B. Group B was the retrospective cohort of 154 patients with non-variceal UGIB who underwent emergency endoscopic hemostasis and in whom antithrombotic agents were subsequently discontinued [11].

Sample size calculation and statistical analysis
The sample size of the non-inferiority study was calculated based on the rate of rebleeding within a month. The rebleeding rate after emergency endoscopic hemostasis in patients with non-variceal UGIB taking antithrombotic agents was 9.7% (15/154 patients) in our previous study [11]. We also considered other studies that described bleeding rates after gastric ESD among patients on antithrombotic treatments (7.0%-21.1%) [21][22][23]. Based on these results, we assumed the overall rebleeding rate in group A to be 20%. We hypothesized that a difference of 10% or less (prespecified non-inferiority margin for this endpoint) in the rebleeding rate between the two groups would constitute a clinically meaningful difference. Assuming a power of 80% and an alpha of 0.05 (one-sided), at least 78 patients (39 patients in each group) would be required in the non-inferiority study. Assuming that 10% of patients could be lost to follow-up, a sample size of 86 patients (43 patients in each group) was planned. The rebleeding rate in group A would be considered non-inferior to that in group B if the 95.0% confidence interval (CI) for the treatment difference was greater than 10.0%.
Treatment outcome and adverse events were compared between groups A and B using propensity score matching analysis. This method was applied to adjust significant differences in the baseline characteristics of the patients and reduce the influence of possible confounding factors [24]. The two groups were matched at a 1:1 ratio (40 patients in each group) with the adjustment for nine covariates (age, gender, H. pylori infection, number of agents (antithrombotic, antiplatelet, and anticoagulant) as well as ulcer characteristics (location, number, and size) to minimize inherent bias. Categorical data were expressed as a number (percentage), and the χ2 test was used to identify differences between the two groups. Numerical data for a normal distribution were expressed as the mean ± the standard deviation, and Student's t-test was used to determine differences between the two groups. And numerical data for a skewed distribution were expressed as median interquartile range and the Mann-Whitney U test was used. Report levels of significance for all comparisons made, whether significant or not, with P-values or confidence intervals. A P-value of <0.05 was considered to be statistically significant for each test. All statistical analyses were performed with JMP, version 13.0.0 (SAS Institute Inc., Cary, NC, USA).

Clinical characteristics of the prospective cohort
Forty-three patients who resumed antithrombotic agents without cessation periods were allocated to group A. The characteristics of patients in group A are listed in Table 1. The mean age of the patients was 77.8 years, and 28 patients (65.1%) were male. The rate of H. pylori infection was 46.5%. Cardiovascular disease was the most common comorbidity; it had been diagnosed in 23 patients. Table 2 summarizes the details concerning antithrombotic agents among the patients. Thirty-two patients (74.4%) took a single antithrombotic agent and 11 patients (25.6%) took multiple antithrombotic agents. The most common antiplatelet agent was aspirin (17 patients), and edoxaban (six patients), one of the direct oral anticoagulants (DOACs), was the most frequently prescribed anticoagulant agent.

Endoscopic findings and clinical outcome of the prospective cohort
The endoscopic findings of patients with non-variceal UGIB are shown in Table 3. Of the 41 patients with peptic ulcer disease, 34 patients complained of bleeding gastric ulcers, among whom multiple gastric ulcers were found in 15 patients. A large ulcer (>11 mm in diameter) was found in 21 patients, and spurting gastroduodenal bleeding was endoscopically identified in seven patients. To achieve endoscopic hemostasis, the soft coagulation method was applied to all but one patient, and 27 procedures (62.8%) were performed by trainees. Consequently, endoscopic hemostasis was successfully carried out in all of the 43 patients.
The mean hemoglobin level on admission was 8.0 ± 2.6 g/dL and blood transfusion was required with a mean volume of 3.2 units. After endoscopic hemostasis, all patients were treated with fasting and hospitalized for a mean period of 2.2 ± 0.7 d and 9.1 ± 5.5 d, respectively. Rebleeding occurred in three patients (7.0%); however, neither major thromboembolic events nor fatalities were recorded during the follow-up period. Table 4 compares the clinical characteristics between groups A and B before and after propensity score matching. Before propensity score matching, the mean age  Table 5 compares the treatment outcomes after propensity score matching between the two groups. Because group B comprised a retrospective cohort of patients who transiently discontinued antithrombotic agents after emergency endoscopic hemostasis, the cessation period of antithrombotic agents was significantly shorter in group A than in group B (0 vs. 6.5 d; P < 0.001). Although the soft coagulation method was more frequently applied in group A than in group B (97.5% vs. 60.0%, P < 0.001), endoscopic hemostasis was successfully achieved in all patients of both groups. The hospitalization periods were significantly shorter in group A than in group B (8.6 ± 5.2 vs. 14.4 ± 7.1 d; P < 0.001).

Comparison of clinical outcomes between groups A and B by propensity score matching
Rebleeding within one month occurred in two (5.0%) of 40 patients in group A and six (15.0%) of 40 patients in group B. The treatment difference was calculated as 10.0% (95.0% CI: −4.2 to 23.2), within an a-priori −10% non-inferiority margin. Therefore, the rebleeding risk in the non-cessation method of antithrombotic agents after emergency endoscopic hemostasis was not considered to be inferior to the transient discontinuation method in patients with non-variceal UGIB (Figure 1). Neither thromboembolic events nor fatality occurred during the follow-up period.

Discussion
In this super-aging society, both antiplatelet and anticoagulant agents have been increasingly used for the prophylaxis and treatment of vascular disease, especially cerebrovascular and cardiovascular disease. However, this practice has also led to an increase in the incidence of and morbidity associated with UGIB [25][26][27][28]. In addition, because the combination therapy of antithrombotic agents has become more prevalent [11], and because these medications may be associated with a higher risk of UGIB [29], medical management in patients with UGIB on antithrombotic agents has been a matter of recent debate.
The JGES guideline published in 2012 drastically changed the management of gastroenterological endoscopy in patients undergoing antithrombotic treatment [14].
The guideline was based on the fact that the clinical outcome of thromboembolic events caused by the transient discontinuation of antithrombotic agents during the periendoscopic period were more severe compared with those of bleeding events among patients treated with the medication [30]. In fact, we also managed two patients who developed severe thromboembolic events after cessation of antithrombotic agents because of UGIB [11]. In contrast, repeat rebleeding in patients with UGIB after endoscopic hemostasis is troublesome for endoscopists, and some reports suggested a possible higher risk of rebleeding in patients with UGIB on antithrombotic agents [31,32]. Furthermore, the clinical management of antithrombotic agents after invasive gastrointestinal endoscopy remains uncertain. We therefore compared the risk of rebleeding and safety of non-variceal UGIB between patients on antithrombotic agents with our non-cessation method and those treated in accordance with the JGES guidelines.
In the present study, the overall rebleeding rate within a month was 7.0% in patients with non-variceal UGIB who continued antithrombotic agents after endoscopic hemostasis. The rebleeding rate in our previous cohort, classified as group B in the present study, in whom antithrombotic agents were transiently discontinued, was 9.7% (15/154 patients) [11]. That value was almost equivalent to the rebleeding rates in other previous studies (7.0% to 17.8%) [32,33,34]. Consequently, the difference in the rebleeding rate between groups A and B after propensity score matching was demonstrated to be within the non-inferiority margin. We therefore considered that our non-cessation method of antithrombotic agents did not lead to an increased risk of rebleeding after emergency endoscopic hemostasis in patients with non-variceal UGIB.
The rebleeding rate after endoscopic hemostasis can be influenced by the hemostatic procedures applied. In the present study, the rate of rebleeding was numerically but not significantly higher in group B than in group A after propensity score matching. Because the soft coagulation method is a useful hemostatic procedure during ESD [20,[35][36][37][38], ESD has also become more prevalent for endoscopic hemostasis in patients with non-variceal UGIB in Japan [16,39]. In addition, because our previous study demonstrated that soft coagulation was significantly superior to hemoclipping for endoscopic hemostasis in such patients [16], we have preferentially chosen the soft coagulation method over hemoclipping. As a result, soft coagulation was more frequently applied in group A than in group B (97.5% in group A vs. 60.0% in group B, P < 0.001). However, it seems unlikely that the difference in the procedure affected the rebleeding rate because the rebleeding rates by hemoclipping and by soft coagulation were similar in group B (14.3% vs. 16.7%, respectively).

No patient experienced major complications (bleeding or thromboembolism)
in the present study. However, we previously reported adverse events in two patients who experienced cerebral infarction during the cessation period of antithrombotic agents [11]. Another report also described six thromboembolic events occurring in 36 patients with non-variceal UGIB who discontinued warfarin as well as one myocardial infarction in 41 patients with non-variceal UGIB who discontinued aspirin [32]. A meta-analysis also demonstrated that the resumption of warfarin following interruption because of GI bleeding was significantly associated with a reduction in thromboembolic events (hazard ratio [HR]: 0.68, 95% CI: 0.52-0.88) and in mortality (HR: 0.76, 95% CI: 0.66-0.88) [40]. These results indicated that we should be strictly aware of the potential for thromboembolic events by the interruption of antithrombotic agents after UGIB. Together with the fact that no major bleeding adverse events occurred in group A of the present study, the non-cessation method after emergency endoscopic hemostasis seems to be appropriate for the management of patients with non-variceal UGIB on antithrombotic treatment. While no significant difference in rebleeding and thromboembolic adverse events were identified in comparing groups A and B, the hospitalization period was significantly shorter in the former than in the latter. Such a difference might be partly attributed to the numerically higher rate of rebleeding in group B. However, this apparent advantage of the non-cessation method of antithrombotic agents after endoscopic hemostasis should be considered cautiously because warfarin requires close monitoring of the prothrombin time and international normalized ratio when resuming.
Warfarin was used more frequently as anticoagulant therapy in group B (nine of nine patients, 100%) than in group A (two of 11 patients, 18.2%). Because DOACs have recently become the mainstay of anticoagulant therapy, further study is necessary to determine the difference in hospitalization period among patients with non-variceal UGIB taking DOACs.
Our present study has two limitations. First, the number of subjects in the prospective cohort was small because of the study design. Therefore, the low rebleeding rate in the prospective cohort can be a statistical type II error. Second, the retrospective cohort of our previous study was used to compare clinical outcome and adverse events between patients with and without a cessation period of antithrombotic agents after endoscopic hemostasis. While propensity score matching analysis was applied to minimize the difference in clinical characteristics between the two groups, further prospective study with a larger number of subjects should be done to validate the results of the present study.

Conclusions
This multicenter pilot study demonstrated that continuing antithrombotic agents without a cessation period was acceptable after emergency endoscopic hemostasis in patients with non-variceal UGIB and could shorten the hospitalization period.

Acknowledgments
The authors thank Dr. Ayako Takamori, Clinical Research Center, Saga University Hospital, for statistical advice. The authors also thank Andrea Baird, MD, from Edanz Group (https://en-author-services.edanzgroup.com/) for editing a draft of this manuscript.

Funding
No specific grants from any funding agencies in the public, commercial, or not-for-profit sectors were received for this research.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Consent for publication
Not applicable.