Langenbecks Arch Surg (2003) 387:327–336 DOI 10.1007/s00423-002-0329-3 Michael Imhof Christian Ohmann Hans-Dietrich Röher Holger Glutig Received: 18 February 2002 Accepted: 20 September 2002 Published online: 13 December 2002 © Springer-Verlag 2002 The authors wrote this article on behalf of the DUESUC study group. The remaining members of the group are: J.-R. Delgadillo, B. May, J. Borski (Krankenhaus, Benrath); R. Mewes, R. Jurisch (Dominikus Krankenhaus); S. El Mohammed, (Ev. Krankenhaus Wülfrath); H. Zeller, T. Kaufmann, (Diakoniewerk Kaiserswerth); C. Leibl, U. Martin, E.-G. Wenzel (Augusta Krankenhaus); A. Müller, C. Ruppert (Krankenhaus Gerresheim); P.R.Verreet (Heinrich-Heine-Universität); J. Haager, J. Heise (Städt. Krankenanstalten Krefeld); P. Goesch, S. Tho Pesch (Krankenhaus Maria Hilf, Krefeld); H. Stallkamp (Städtisches Krankenhaus Leverkusen); K.P. Thon, H. Stöltzing (Robert Bosch Krankenhaus Stuttgart) M. Imhof Department of General and Trauma Surgery, St.-Vinzenz Hospital, Düsseldorf, Germany C. Ohmann (✉) Coordination Centre for Clinical Trials, Heinrich Heine University, Moorenstrasse 5, 40225 Düsseldorf, Germany e-mail: ohmannch@uni-duesseldorf.de Tel.: +49-211-8119701 Fax: +49-211-8119702 H.-D. Röher · H. Glutig Department of General and Trauma Surgery, Heinrich Heine University, Düsseldorf, Germany O R I G I N A L A RT I C L E Endoscopic versus operative treatment in high-risk ulcer bleeding patients – results of a randomised study Abstract Background and aims: The treatment of peptic ulcer bleeding is still a matter of controversy in high-risk patients with arterial bleeding or a visible vessel. Patients with recurrent bleeding during hospital stay carry an increased risk of death. Therapeutic concepts using early elective operation compete with solitary endoscopic treatment. Numerous prospective studies have contrasted comparable improvements for the different therapeutic regimens but there is still a lack of a randomised comparison. Patients and methods: We initiated a multicentre randomised clinical trial comparing endoscopic fibrin glue injection with early elective operation in peptic ulcer patients with arterial bleeding or a visible vessel ≥2 mm. After initial endoscopic control of bleeding, patients were randomised to repeated fibrin glue injection or early elective operation. Outcome criteria were recurrent bleeding and death. The study was terminated after a planned interim analysis. Results: Due to strict inclusion and exclusion criteria 61 patients were randomised and 55 patients could be included in the perprotocol analysis, 23 in the early elective operation group and 32 in the patient group with endoscopic therapy. The type of surgery in the early elective operation group was usually gastric resection (79%). Recurrent bleeding occurred in 50% of the endoscopically treated patient group, and in the operative group in one patient (relative risk: 11.5; 95% CI: 1.6 to 80.7). There were no statistically significant differences between the two treatment groups with respect to mortality (relative risk: 0.7, 95% CI: 0.1 to 4.8). Conclusion: Early elective surgery is an effective procedure in bleeding peptic ulcer patients at high risk for re-bleeding. Fibrin glue injection carries a risk for re-bleeding, however, the majority of these re-bleeding episodes can be controlled by reendoscopic treatment, but a subgroup will need an emergency operation with a fatal outcome in individual patients. Keywords Peptic ulcer bleeding · Randomised trial · Surgical treatment · Fibrin glue sealing 328 Table 1 Classification of bleeding activity [3] Bleeding activity Risk Type Ia: active arterial bleeding Type IIa: non-bleeding visible vessel Type Ib: oozing bleeding Type II: signs of recent bleeding (adherent clot, blood or coagula in upper GI tract) Type III: no signs of recent bleeding (haematemesis or melaena in the past 48 h) High Introduction Actively bleeding duodenal or gastric ulcers represent the major and potentially life-threatening complication of peptic ulcer disease, with an average mortality of 10% that may be considerably higher in severely bleeding gastroduodenal ulcers. Despite the introduction of effective medical treatment, and the new pathophysiological understanding of peptic ulcer disease as a chronic infection, no decrease in the incidence of bleeding has been observed so far [1]. One possible reason could be a changing risk profile with more aged and multi-morbid patients [2]. The so-called bleeding activity (Table 1) is still the most important prognostic factor for the final outcome [3]. Patients at high risk for persistent or recurrent bleeding are those with spurting arterial bleeding or a visible vessel at emergency endoscopy [3, 4]. The treatment of patients at low risk (oozing bleeding, signs of recent bleeding and ulcers with no signs of recent bleeding) can be managed endoscopically and medically, and patients do well in any event. For ulcer bleeding the standard procedure consists of diagnostic emergency endoscopy and endoscopic treatment based on the bleeding activity. Patients with recurrent bleeding during hospital stay carry an increased risk of death. In studies, up to 50% of these patients have been operated on, with a mortality of up to 25% [5, 6, 7, 8]. For that reason, an early elective operation with definitive haemostasis has been recommended for patients with a high risk of further bleeding [9]. In some studies excellent results have been reported for this approach; other authors, however, favour endoscopic treatment. Unfortunately, no controlled trial was available that compared effective endoscopic treatment and early elective operation in high-risk ulcer bleeding patients. Therefore, we performed a randomised controlled clinical trial comparing early elective operation with endoscopic treatment in this patient group. Materials and methods The Duesseldorf study of ulcer complication II (DUESUC II) was performed as a randomised multicentre and interdisciplinary study with ten participating hospitals (surgery and internal medicine), between 16 September 1991 and 31 December 1995 (see extended abstract, [10]). Necessary assumption for the participation of a Low hospital was the inclusion of the surgical and internal department and the guarantee of a 24-h on-duty endoscopy staff. Included were high risk peptic ulcer patients with A. Spurting arterial bleeding or a visible vessel equivalent to or greater than 2 mm in diameter at emergency endoscopy that had to be performed within 6 h of admission and B. Endoscopic control of active bleeding by fibrin glue injection. The size of the visible vessel was estimated by the biopsy forceps or the sclerosing needle. Patients who developed ulcer bleeding during hospitalisation and/or under defined severe stress conditions (for example, severe polytrauma, severe burns, renal insufficiency), patients with Dieulafoy’s ulcer, Barrett ulcer, patients with previous stomach resections, age <18 years, with malignant growth in upper GI tract, portal hypertension with oesophageal varices, pregnancy, moribund (ASA 5) or serious systemic disease (ASA 4), Marcumar patients with a Quick test result<30% and patients with no informed consent were excluded. For the definition of the bleeding activity the Forrest classification was used [3, 4] (Table 1). Outcome criteria were recurrent haemorrhage and death (during hospital stay). After initial endoscopic control of bleeding by fibrin glue the patients were randomised to: ● ● Group A: early elective operation or Group B: fibrin glue treatment If randomised to early elective operation, patients with duodenal ulcers had to be operated on within 8 h, those with gastric ulcers within 24 h; the type of operation was recommended, but not determined. If patients were randomised to the endoscopic treatment group, endoscopic control was performed within 24 h with repeated fibrin glue treatment if necessary (signs of bleeding). Fibrin glue sealing was performed again in cases of recurrent bleeding. In cases of further recurrent bleeding individual treatment could be performed. All patients in the endoscopic group and all patients with non-resective operations had adjuvant medical treatment by omeprazole. Eradication was not intended in the study protocol. A balanced block randomisation for patients (block size: 8) was performed with stratification according to centre. The allocation schedule was generated by the Theoretical Surgery Unit not involved in patient treatment, using random numbers generated by Ciba-Geigy. Central telephone randomisation was the standard procedure except in cases where this was not possible (prepared sealed envelopes in the study centre). Assignment was performed after initial endoscopic control of bleeding by fibrin glue treatment. Blinding of treatment was not possible. A target sample size of 120 was projected, based on the outcome criteria: recurrent bleeding, a bleeding rate of 10% for the operative group, a clinically important difference of 20% and α=0.05, β=0.20. According to protocol, an interim analysis was scheduled after 60 patients (adjusted α=0.029). ‘Intention-to-treat’ and ‘per-protocol’ analyses were performed. The outcome criteria: recurrent bleeding and death during hospital stay, were tested with the Fisher exact test. Findings for re-bleeding were fresh haematemesis for melaena, fresh blood in the nasogastric tube (mandating within the first 24 h