Randomised comparison between adrenaline injection alone and adrenaline injection plus heat probe treatment for actively bleeding ulcersBMJ 1997; 314 doi: https://doi.org/10.1136/bmj.314.7090.1307 (Published 03 May 1997) Cite this as: BMJ 1997;314:1307
- Sydney S C Chung, professora,
- James Y W Lau, senior medical officera,
- Joseph J Y Sung, readerb,
- Angus C W Chan, lecturera,
- C W Lai, senior medical officerb,
- Enders K W Ng, senior medical officera,
- Francis K L Chan, senior medical officerb,
- M Y Yung, research assistanta,
- Arthur K C Li, chairman and professora
- a Department of Surgery Prince of Wales Hospital Chinese University of Hong Kong Shatin New Territories Hong Kong
- b Department of Medicine Prince of Wales Hospital Chinese University of Hong Kong Shatin New Territories Hong Kong
- Correspondence to: Dr Li
- Accepted 24 January 1997
Objective: To compare endoscopic adrenaline injection alone and adrenaline injection plus heat probe for the treatment of actively bleeding peptic ulcers.
Design: Randomised prospective study of patients admitted with actively bleeding peptic ulcers.
Setting: One university hospital.
Subjects: 276 patients with actively bleeding ulcers detected by endoscopy within 24 hours of admission: 136 patients were randomised to endoscopic adrenaline injection alone and 140 to adrenaline injection plus heat probe treatment.
Main outcome measures: Initial endoscopic haemostasis; clinical rebleeding; requirement for operation; requirement for blood transfusion; hospital stay, ulcer healing at four weeks; and mortality in hospital.
Results: Initial haemostasis was achieved in 131/134 patients (98%) who received adrenaline injection alone and 135/136 patients (99%) who received additional heat probe treatment (P = 0.33). Outcome as measured by clinical rebleeding (12 v 5), requirement for emergency operation (14 v 8), blood transfusion (2 v 3 units), hospital stay (4 v 4 days), ulcer healing at four weeks (79.1% v 74%), and in hospital mortality (7 v 8) were not significantly different in the two groups. In the subgroup of patients with spurting haemorrhage 8/27 (29.6%; 14.5% to 50.3%) patients from the adrenaline injection alone group and 2/31 (6.5%; 1.1% to 22.9%) patients from the dual treatment group required operative intervention. The relative risk of this was lower in the dual treatment group (0.17; 0.03 to 0.87). Hospital stay was significantly shorter in the dual treatment group than the adrenaline injection alone group (4 v 6 days, P = 0.01).
Conclusion: The addition of heat probe treatment after endoscopic adrenaline injection confers an advantage in ulcers with spurting haemorrhage.
Endoscopic injection of adrenaline alone is effective in stopping bleeding peptic ulcers
Further bleeding after adrenaline injection alone, however, occurs in 15-20% of patients, and the addition of heat probe thermocoagulation may improve permanent haemostasis and therefore patient outcome
When compared with adrenaline injection alone the dual treatment significantly reduced the requirement for operative intervention and the length of hospital stay in the subgroup of patients with spurting ulcer haemorrhage
In the endoscopic treatment of spurting ulcer haemorrhage heat probe thermocoagulation should be added after adrenaline injection
A decade ago surgery was the only effective method of controlling bleeding ulcers. With the advent of effective means of endoscopic haemostasis, endoscopic treatment has emerged as the first line treatment of ulcer haemorrhage. Many methods, such as laser, contact thermal probes, or injection therapy, have been proved to be clinically useful.1 2 3 4 5 6 7 8 9 10 Which one of these methods is superior remains controversial.
Contact thermal probes use the principle of captive coagulation: the vessel wall is pressed together by the probe before the probe is activated. The vessel is sealed by a combination of pressure and heat. Heat probe treatment has been shown to improve clinical outcome in patients with actively bleeding ulcers and high risk stigmata. For optimum effect, the larger (10 French gauge, 3.2 mm) probe has to be used. A larger, less manoeuvrable therapeutic endoscope with a 3.7 mm instrument channel is therefore required. The successful application of a heat probe also demands accurate targeting and firm tamponade of the bleeding vessel. In the presence of flowing blood this is not always possible.
We have previously compared heat probe treatment with adrenaline injection in the treatment of actively bleeding ulcers.11 The success in stopping active bleeding at the time of endoscopy was lower with the heat probe compared with adrenaline injection (83% v 96%). Adrenaline injection is highly effective in stopping active ulcer bleeding. In our previous studies with this technique the success rate in stopping active bleeding ranged from 93% to 100%.7 11 12 13 Adrenaline, however, does not induce permanent thrombosis in blood vessels. About 13% to 20% of patients develop further bleeding, which often results in emergency surgery. We attempted to improve the results by the additional injection of a sclerosant. Neither the additional injection of 3% sodium tetradecyl sulphate nor absolute alcohol reduced the incidence of rebleeding.12 13 The dual treatment of adrenaline injection followed by heat probe application was therefore theoretically appealing. Active bleeding is first controlled by injection so that the eroded vessel can be clearly seen, allowing accurate placement of the heat probe. Cessation of active bleeding also means that heat energy delivered by the probe is not carried away by the flowing blood.
We report the results of a randomised prospective trial that compared adrenaline injection alone with adrenaline injection followed by heat probe treatment in actively bleeding ulcers. We examined whether the dual treatment improves permanent haemostasis and would therefore reduce mortality in patients actively bleeding from peptic ulcers.
Patients and methods
The Prince of Wales Hospital in Hong Kong is a teaching hospital serving a population of about 1.2 million. Patients admitted with acute upper gastrointestinal bleeding underwent endoscopy within 24 hours of admission. Patients with fresh haematemesis and those who were haemodynamically unstable underwent endoscopy after initial resuscitation. In all patients informed consent was obtained for possible endoscopic haemostasis. Gastric lavage was used only if a good view of the bleeding point could not be obtained, in which case an overtube was passed. Blood covering the ulcer was washed away with a “Waterpik.”14 Patients with actively bleeding ulcers (spurting or oozing haemorrhage) identified at the time of endoscopy were recruited into the present study. Patients with non-bleeding visible vessels, adherent blood clots, clean based ulcers, or ulcers with contact bleeding only were excluded.
Approval was obtained from the local ethics committee before we started the trial. Patients were then randomised to receive either adrenaline injection alone or adrenaline injection followed by heat probe treatment. Randomisations took place at the time of endoscopy when actively bleeding ulcers were seen. The endoscopy nurse then opened a sealed envelope containing the treatment option, which had previously been determined by a random number list generated by a computer. Treatment was concealed from the endoscopist when the patient was admitted into the trial.
Adrenaline in 1:10 000 dilution was injected in 0.5-1 ml aliquots into and around the bleeding point until the bleeding was controlled. The group randomised to dual treatment received additional captive coagulation with an Olympus heat probe unit (Olympus Optical, Tokyo). The 10 French gauge heat probe (Olympus CD-1OZ) was used to tamponade the bleeding point firmly for three pulses of 30 J at any one site. The end point of treatment with the heat probe was defined as flattening or cavitation of the bleeding point. If a regular diagnostic endoscope was used initially, a dual channel therapeutic endoscope (2T-10 or 2T-200; Olympus, Japan) was used for the heat probe treatment usually after the initial control of bleeding by adrenaline injection.
The patients were returned to the surgical gastroenterology ward after treatment. They were managed by surgeons blind to the treatment option. Blood transfusion was given to maintain the haemoglobin concentration at around 100 g/l. All patients underwent endoscopy 24 hours later, and a further injection of adrenaline alone was given if active bleeding was again seen. Heat probes were not used at the second endoscopy because of the risk of perforation.11 Criteria for emergency surgery were drawn up and strictly adhered to. These were arterial bleeding not controlled at the time of endoscopy; clinical rebleeding defined as fresh haematemesis or melaena associated with tachycardia (pulse rate >110 beats/min) or hypotension (systolic blood pressure < 90 mm Hg); or total transfusion exceeding 8 units of blood to maintain a haemoglobin concentration of around 100 g/l. Histamine receptor antagonists, omeprazole, or triple treatment for eradication of Helicobacter pylori were given on discharge, and endoscopy was carried out four weeks later to assess ulcer healing. At follow up endoscopy, endoscopists were blind to the initial treatment option.
The outcome assessments (decided a priori) compared between the two treatment groups were initial endoscopic haemostasis, rebleeding, requirement for surgery, requirement for blood transfusion, length of hospital stay, and mortality in hospital.
From our previous trials, rebleeding after adrenaline injection alone varied between 15% and 20%. No such data existed for the dual treatment of adrenaline injection plus heat probe treatment. If we assumed a rate of rebleeding of 5-10%, a sample size between 176 and 438 would be required to achieve a statistical power of 80% at 5% type I error. An interim analysis at around 220 enrollments was planned, and significance was taken at 3% type I error. We used Student's t test for comparisons of continuous variables, two tailed Pearson's χ2 test or Fisher's exact test for comparisons of categorical variables, and non-parametric Mann-Whitney U test for comparisons of requirements for blood transfusions and length of hospital stay.
Between 1 September 1992 and 7 June 1994 a total of 3558 patients were admitted to the Prince of Wales Hospital with upper gastrointestinal bleeding; 2170 were bleeding from peptic ulcers. Among them, 303 patients were actively bleeding at the time of endoscopy. Twenty seven patients were not randomised; in three the source of bleeding could not be identified because of torrential bleeding, and one patient was not cooperative at time of endoscopy. Heat probe treatment was not possible in 23 patients because the probe unit was out of order from 24 November 1992 to 1 January 1993. A total of 276 patients were recruited into the study. They were randomised to receive either endoscopic injection of adrenaline alone (136) or adrenaline injection plus heat probe treatment (140). Six patients were excluded after randomisation (2 v 4) as their ulcers were subsequently found to be malignant. The two groups were matched in patients' characteristics and severity of bleeding (table 1). There were no significant differences in the age and sex distribution and the proportion of duodenal, gastric, and anastomotic ulcers.
Initial success in endoscopic haemostasis was comparable in the two groups (98% v 99%; odds ratio 3.09; 95% confidence interval 0.32 to 30.12). Primary haemostasis was not possible in three patients from the adrenaline injection alone group and one patient from the dual treatment group. These four patients went directly to surgery, and one subsequently died. At the second endoscopy 13 from the adrenaline injection alone group and nine from the dual treatment group required further treatment for endoscopic rebleeding. Two patients from the injection alone group and one patient from the dual treatment group went directly to surgery as bleeding could not be controlled endoscopically. Further treatments with adrenaline alone were successful in the remaining patients.
In the entire group 12 (9.0%; 4.2% to 13.8%) from the adrenaline injection alone group and five (3.7%; 1.4% to 8.8%) from the dual treatment group had clinical evidence of rebleeding (odds ratio 0.39; 0.13 to 1.13). The difference, however, was not significant. Emergency operations were performed in 14 patients (10.4%; 5.2% to 15.6%) in the adrenaline injection alone group and eight (5.9%; 2.8% to 11.7%) in the dual treatment group (0.54; 0.22 to 1.32). The total number of blood transfusions, length of hospital stay, ulcer healing at four weeks, and mortality in hospital were not significantly different in either group (table 2).
In patients with spurting haemorrhage (table 3) six (22.2%; 9.4% to 42.7%) assigned to adrenaline injection alone rebled clinically with circulatory instability or fresh haematemesis whereas only two (6.5%; 1.1% to 22.9%) from the dual treatment group rebled (0.02; 0.05 to 1.33). Eight patients (29.6%; 14.5% to 50.3%) from the adrenaline injection alone group underwent surgery: six for clinical rebleeding and two because they required more than 8 units of blood. In the dual treatment group two (6.5%; 1.1% to 22.9%) with clinical rebleeding needed surgery. The relative risk of the need for operation was lower in the dual treatment group (0.17; 0.03 to 0.87). Length of hospital stay in the dual treatment group was significantly shorter than that in the injection alone group (4 v 6 days; P = 0.01, Mann-Whitney U test). Requirement for blood transfusion in the dual treatment group was arithmetically less than that in the injection alone group, although this did not reach significance (median 4 v 5 units; P = 0.06, Mann-Whitney U test).
In patients with oozing haemorrhage no difference between two groups was observed in their clinical outcome (table 4). There were two perforations related to the heat probe in the dual treatment group. No complications related to treatment were seen in the adrenaline alone group.
Only 91/134 patients (67.9%) from the adrenaline injection alone group and 96/136 patients (70.6%) from the dual treatment group returned for follow up endoscopy at four week. Ulcer healing was documented in 72/91 patients (79.1%) in the adrenaline injection alone group and 71/96 patients (74%) in the dual treatment group (P = 0.41, Pearson's χ2 test). When we had enrolled 220 patients we carried out an interim analysis while we continued to randomise eligible patients. As an entire group the difference in outcome between two treatment groups was small. In the subgroup of patients with spurting haemorrhage the observed difference between the two treatment groups in their requirement for operation was associated with a statistical power of 27%. We estimated that we would need a sample size of 180 and a further four years of recruitment to achieve a statistical power of 80% at 5% type I error. As it was unlikely that we could continue for such a period we terminated the trial with the enrollment number at 276 and conducted a final analysis.
In recent years many randomised clinical trials on the endoscopic treatment of bleeding ulcers have been reported.1 2 3 5 6 7 8 10 Most trials that compared endoscopic treatment against standard medical treatment reported improvement in patient outcome as measured by reduced blood transfusion, decreased requirement for surgical intervention, and reduced length of hospital stay. An impact on mortality is more difficult to show as mortality in patients with bleeding ulcers is less than 10% in most series and the number of patients in individual trials is small. Recent meta-analyses, however, showed reduction in mortality with endoscopic haemostasis.15 16 As differences in efficacy among different endoscopic devices are likely to be small, comparative studies of different endoscopic treatments have in general failed to show any superiority of one technique over another.11 17 18 19
Inclusion of patients at high risk
In up to 80% of patients with bleeding ulcers bleeding stops spontaneously. Endoscopic treatment should be targeted to those who are at high risk of developing further bleeding. In conducting clinical trials on endoscopic ulcer haemostasis, it is important to include only patients at high risks of recurrent haemorrhage so that significant results in high risks groups are not overshadowed by patients who are unlikely to rebleed.20 The National Institutes of Health consensus conference recommended that only ulcers with active bleeding or visible vessels should be treated endoscopically.21 Recent evidence indicates that patients with ulcers with an adherent clot may also constitute a high risk group.22 The endoscopic diagnosis of stigmata of recent haemorrhage, however, depends on the observer. The agreement between observers in the interpretation of visible vessels is poor even among international experts with interests in endoscopic ulcer haemostasis.23 For this reason, we included only patients with actively bleeding ulcers in the present study.
Attempts at improving permanent haemostasis
We have successfully used adrenaline injection for over 10 years in patients with actively bleeding peptic ulcers. Haemostasis can be achieved in most cases. Considerable rebleeding that required surgical intervention occurred in 15% of patients. From a technical stand point dual treatment with adrenaline injection followed by heat probe treatment is attractive. The spurting haemorrhage is first controlled by injection, which does not need to be accurately targeted as injection close to the bleeding point will suffice to control bleeding. As submucosally injected adrenaline does not damage tissue or have appreciable systemic effects, large volumes can be used without fear of complications.24 Once the active bleeding is controlled the heat probe can be targeted on to the bleeding vessel to apply firm tamponade and captive coagulation. In our previous study that compared adrenaline injection with heat probe treatment the control of active haemorrhage was inferior with the heat probe principally because of problems of accurate placement of the probe in ulcers with active bleeding. Rutgeerts et al showed that the combined use of adrenaline injection and photocoagulation by neodymium: yttrium-aluminium-garnet laser was more effective than adrenaline injection alone in achieving permanent haemostasis.25 A subsequent randomised trial confirmed that the additional laser treatment provided a marginal improvement in primary haemostasis.26 Although the additional injection of a sclerosant shares the same theoretical appeal, we have been unable to show that the additional injection of either sodium tetradecyl sulphate12 or absolute alcohol13 confers any advantage in improving permanent haemostasis.
Combining injection and contact thermocoagulation
As in our previous studies, practically all active bleeding was controlled by adrenaline injection at the time of endoscopy.4 7 11 12 13 As an entire group, although there was a trend favouring dual treatment, clinical outcomes in the two groups were not significantly different. When we analysed data for patients with arterial spurting and non-pulsatile oozing separately, however, significant differences were apparent. In patients with arterial spurting the need for emergency surgery decreased from 29.6% to 6.5% in the dual treatment group. The length of stay in hospital was also significantly reduced (P = 0.01). In the non-pulsatile group (table 4), however, the clinical outcomes were identical in the two treatment arms.
Arterial spurting at the time of endoscopy is a grave prognostic sign. In one of our previous studies seven out of 10 patients with arterial spurting required emergency surgery if no endoscopic treatment was carried out.7 Although adrenaline injection is highly effective in controlling active bleeding even in spurting haemorrhage, it does not induce permanent vessel thrombosis. The exact mechanism of action is probably a combination of vasoconstriction, tissue tamponade, and platelet aggregation.27 28 29 A thrombus then seals the vent in the eroded artery. In ulcers in which a large vessel is eroded rebleeding may occur when the thrombus dislodges. Our results indicate that captive coagulation of the bleeding vessel, even after haemostasis by adrenaline injection, reduces further bleeding and the requirement for an operation in patients presenting with arterial spurters. Compared with ulcers with arterial spurting, ulcers with non-pulsatile oozing are probably bleeding from smaller blood vessels. The chance of dislodgment of the clot and recurrent haemorrhage is therefore smaller, and the advantage of additional captive coagulation not apparent. Ulcers with oozing haemorrhage constitute a heterogenous group. Steele proposes that oozing should be further classified into oozing without a visible vessel, oozing with a visible vessel, and oozing under an adherent clot (personal communication).
We conclude that in the endoscopic treatment of ulcers with spurting haemorrhage the addition of heat probe thermocoagulation to adrenaline injection seems to reduce rebleeding and the need for subsequent emergency surgery and should be carried out even if bleeding is initially controlled by adrenaline injection.
Funding: Croucher Foundation.
Conflict of interest: None.