In the management of patients with acute upper gastrointestinal bleeding from non-variceal causes, endoscopic treatment and acid suppression are now the standard of care. Current endoscopic treatment in the form of either thermo-coagulation or clipping to the bleeding arteries is highly efficacious in the stopping bleeding. Unfortunately in 5 to 10% of patients, bleeding cannot be controlled during index endoscopy or recurs after initial hemostasis. These patients are often elderly with significant co-morbidities. Their bleeding lesions are large eroding into major sub-serosal arteries. In the few who need surgical salvage, mortality increases to around 30%. The Over-the-scope-Clip (OTSC) is a device, which allows endoscopists to capture a large amount of tissue and compress on the bleeding artery. The OTSC also has a high retention rate. Recurrent bleeding with the use of standard hemo-clips can occur because of their low retention rate. We reported the use of OTSC with a high success rate in a case series of patients with refractory bleeding after standard endoscopic treatment. We have also used OTSC in the treatment of bleeding from pseudo-aneurysm arising from large eroded arteries in ulcer base. A multicenter randomized controlled trial that compares OTSC to standard endoscopic treatment in the endoscopic treatment of refractory bleeding lesions has just been completed. The use of OTSC has been shown to be superior in achieving hemostatic control and reducing further bleeding. In this proposed randomized controlled trial, we would test the hypothesis that the use of OTSC, when used as the first or primary treatment, is superior to standard treatment in achieving hemostasis and thereby improve patients' outcomes.

Endoscopic therapy has improved outcomes to patients with acute non-variceal upper gastrointestinal bleeding. Acid suppression therapy as an adjunct to endoscopic therapy further reduces recurrent bleeding and interventions. There remains a small subgroup of patients who continue to bleed or re-bleed after initial endoscopic hemostasis. These patients are often elderly with significant comorbid illnesses. In the 2007 National United Kingdom Audit, mortality was 30% in those after salvage surgery for refractory bleeding [1]. The current standard in endoscopic treatment is the use of either hemo-clips or thermal coagulation with or without pre-injection with diluted adrenaline. The two modalities are similar in their efficacies [2]. Their use is associated with a small but significant failure rate. Anatomically, the failed cases are usually larger deep ulcers with erosion into thick subserosal arteries [3]. In an ex vivo model, control of bleeding becomes inconsistent in arteries > 2mm in diameter in size with 3.2 mm contact thermal devices [4]. The investigators have few reports on in vivo data. Using thin barium angiography in 27 gastrectomy specimens from patients who underwent surgery for control of bleeding from their gastric ulcers, Swain and colleagues found vessels of varying sizes and disposition [5]. It was felt that artery beyond 1 mm would be difficult to control during endoscopy in a clinical setting. In a series of fatal deaths from bleeding gastroduodenal ulcers from Hong Kong [6], these arteries were around 3 mm in diameter at post-mortem examination.

Over-the-scope clips or the Bear Claws were developed for the purpose of closure of GI luminal defects created at Natural orifice trans-luminal surgery. In clinical practice, OTSC are widely used for closure of GI perforations and anastomotic dehiscence. There have been anecdotal reports over the use of OTSC in the control of refractory bleeding from peptic ulcers [7, 8]. The investigators reported our experience in 9 such patients with successful hemostasis in all of them [9]. Further bleeding occurred in 2 patients. The investigators subsequently illustrated its use in difficult ulcers that failed standard endoscopic treatments; one of them showed a pseudo-aneurysm to a branch of the gastro-duodenal artery [10]. In our hospital, indication for OTSC use has been failure with conventional endoscopic treatments.

The OTSC have several advantages over existing treatments. First, tissue compression on the bleeding artery is critical in control of bleeding. OTSC captures tissue size to that of an endoscope at least 9 mm in diameter. In theory, OTSC can compress arteries of significant size, beyond what can be achieved with standard treatment. Second, clip retention rate with OTSC is extremely and recurrent bleeding from clip loss or dislodgement can be avoided. Third, local complication from their applications is negligible. As a comparison, contact thermal coagulation carries a small risk (around 1%) of perforation.

Investigator group has participated in a multicenter randomized trial [11] that compares OTSC to standard therapies in endoscopic management of refractory non-variceal bleeders. OTSC has been shown to be superior in the control of bleeding (30/33 patients, 91% vs. 15/26 patients 57.6%, P=0.005) and is associated with fewer re-bleeding after hemostasis 6/33, 18% vs. 10/26, 38%, P=0.14).

In the current proposed randomized controlled trial, the investigators compare OTSC and standard therapies (hemo-clipping or thermal coagulation) as the primary endoscopic treatment in the control of bleeding from non-variceal causes in patients who present with acute upper gastrointestinal bleeding. The investigators argue that OTSC should be used as the first treatment given its promise in better bleeding control over other treatment methods. Rebleeding patients are exposed to excessive risks in organ failure and death. Furthermore, management of recurrent bleeding is expensive e.g. cost of angiographic embolization, surgery and intensive care stay.

Investigators hypothesize that OTSC as the first treatment during index endoscopy is superior to standard endoscopic therapy in the overall rate of hemostasis. Knowledge from this clinical trial will enable us to choose the correct treatment for bleeding lesions.

Research plan and methodology Trial Design - A randomized controlled trial The trial will be executed in accordance with Good Clinical Practice guidelines Randomization - Patients with overt signs of acute upper GIB (melena, hematemesis, drop in haemoglobin with or without hypotension) are invited to participate in the trial. Investigators or research nurses approach patients and obtain their written consents. During endoscopy, if they have bleeding lesions suitable for standard endoscopic treatment and OTSC, the endoscopist then requests the next number sealed envelope to be opened by an endoscopy nurse. Patients are randomized to receive either standard endoscopic treatment (contact thermal methods or hemoclips with or without pre injection with diluted epinephrine), or the use of OTSC. Indications for endoscopic treatment are active bleeding (pulsatile or Forrest Ia bleeding, oozing from a visible vessel or a protuberance otherwise referred to as Forrest I b bleeding, or a non bleeding visible vessel or Forrest IIa lesion). Clots overlying bleeding lesions are irrigated or elevated using a cheese wiring technique with a minisnare. If a vessel is unveiled, randomization can then proceed and the assigned treatment is then carried out. The random number list is generated by computer software.

Blinding - no blinding

Participants - Consecutive patients admitted with overt signs of acute upper GIB (melena, hematemesis, drop in hemoglobin with or without hypotension) and documented bleeding (Forrest I) from a non-variceal upper gastrointestinal sourcegastro-duodenal ulcers, Mallory Weiss tear, cancers, Dieulafoy's and other vascular lesions) at endoscopy.

Interventions - Patients are randomly assigned at endoscopy to receive;

Patients are randomized in a 1:1 ratio to receive; endoscopic treatment using OTSC or standard therapies.

No cross over to treatment of the other arm is allowed. A full description of the bleeding lesion and the success of endoscopic therapy are provided and documented in the electronic endoscopy report. It is unrealistic and impossible to completely blind ward staff as a plain abdomen x-ray will readily an OTSC or indeed most hemo-clips.

After successful endoscopy, patients are started on an intravenous infusion with proton pump inhibitor (PPI) for 72 hours. Thereafter, patients go onto receive an oral PPI or standard PPI H. pylori eradication therapy should the patient is infected. It is generally recommended that patients be transfused to a hemoglobin level of around 7 or 8 g/dl. In patients with significant cardiac co-morbidities, a higher transfusion threshold is allowed.

The management of antiplatelet agents and anticoagulation in these patients is based on ASGE guidelines.

Further bleeding is defined by failure to control bleeding during first endoscopy or recurrent bleeding after initial control. Recurrent bleeding is defined by fresh hematemesis, fresh melena or hematochezia with hemodynamic instability (systolic blood pressure lower than 90mmHg, heart rate greater than 110 per minute) and/or drop in hemoglobin > 2g/dl in 24 hours after transfusion to around 8 g/dl. An episode of recurrent bleeding has to be confirmed by repeat endoscopy showing fresh blood in stomach or active bleeding from the previously treated lesion.

Statistics analysis An independent data monitoring and safety panel will be formed by senior physicians from our Clinical Trial Centre at the Chinese University of Hong Kong to adjudicate over the occurrence of primary outcome events.

Analyses are by intention-to-treat and then per protocol principles. The primary outcome between groups over 30 days is compared using a time-to-event type analysis with a log rank test. An adjusted analysis or Cox regression model will be used in the presence of significant difference in the baseline covariates. Initial rate of hemostasis and recurrent bleeding within 30 days are independently reported and compared using a chi-square test. Categorical data in secondary outcome measures are compared between groups again using chi-square test. Parametric and non-parametric data are compared using student's t test and Mann-Whitney U test respectively.

Inclusion Criteria:

• Patients with overt signs of acute upper GIB (melena, hematemesis, drop in hemoglobin with or without hypotension)
• documented bleeding lesions suitable for standard endoscopic treatment during endoscopy

Exclusion Criteria:

• without a full informed consent from the patient or his legally-acceptable representatives
• Age <18 years
• Pregnant
• Lactating women
• Moribund patients not considered for active treatment.