Controlled Study of Different Sclerosing Agents for Coagulation of Cahine Gut Arteries

Transcription

1 GASTROENTEROLOGY 1989;96: Controlled Study of Different Sclerosing Agents for Coagulation of Cahine Gut Arteries GAYLE M. RANDALL, DENNIS M. JENSEN, KENNETH HIRABA Y ASHI, and GUSTAVO A. MACHICADO Medical and Research Services, UCLA Center for the Health Sciences, Center for Ulcer Research and Education (CURE), and West Los Angeles Veterans Administration Medical Center, Los Angeles, California Excellent clinical results have been reported with sclerotherapy for control of nonvariceal gastrointestinal hemorrhage. However, there are few controlled or coniparative data on different sclerosing agents for treatment of lesions with active arterial bleeding or nonbleeding visible vessels. In a controlled, randomized study of canine small bowel arteries our purposes were (a) to evaluate the efficacy for arterial coagulation of six sclerosing agents compared with normal saline control, (b) to compare the resultant tissue injury of agents, and (c) to elucidate the possible mechanisms of arterial coagulation and tissue injury of the agents. The agents evaluated were (a) 98% ethanol, (b) TES, a mixture with final concentration of 1% tetradecyl sulfate, 32% ethanol, and 0.3 normal saline, (c) 1% polidocanol (Ethoxysclerol), (d) 1:10,000 epinephrine, (e) 7.2% hypertonic saline, (0 3.6% hypertonic saline, and (g) 0.9% saline (normal saline control). Agents were injected from the mucosal side of the small bowel into and around the penetrating serosal arteries in the subserosa 1 space. Ethanol and TES were the most effective agents for arterial coagulation. Polidocanol was less effective than ethanol and TES. However, it was the only other agent that induced significant arterial coagulation. Alcohol and TES also caused significant injury in surrounding tissue. The degree of this injury was dependent on the total volume injected. Epinephrine induced significant mucosal damage without significant serosal injury or arterial coagulation. The coagulation and tissue injury effects of hypertonic saline injections were not significantly different from normal saline control. Endoscopic control of active arterial hemorrhage is an important technological advance, particularly for patients at increased risk for emergency surgery. Several endoscopic techniques are reported to be effective and safe for control of severe peptic ulcer hemorrhage. These include argon and yttriumaluminum-garnet lasers, heater probe, bipolar electrocoagulation, and monopolar electrocoagulation. Thermal coagulation of arteries has been extensively studied and some clinically applicable mechanisms such as vessel coaptation have been demonstrated (1-6). Recently, there has been a growing interest in sclerotherapy for the treatment of nonvariceal gastrointestinal (GI) hemorrhage because it is inexpensive, widely available, and relatively easy to use. Although excellent clinical results have been reported with several different sclerosing agents for treatment of nonvariceal GI bleeding (7-22), most of this work was neither controlled nor randomized. Furthermore, there are few comparative data evaluating the efficacy and safety of sclerosing agents for bleeding or nonbleeding arterial lesions. The choice of sclerosing agents for treatment of nonvariceal bleeding remains empirical. In contrast to thermal coagulation, the mechanisms of coagulation and tissue injury for different sclerosing agents have not been extensively studied and are poorly understood. Our study was a controlled, randomized evaluation of sclerosing agents comparing their coagulation effect and associated tissue injury in canine gut arteries. Our purposes were (a) to evaluate the efficacy for arterial coagulation of six different sclerosing agents compared to normal saline control, (b) to compare the resultant tissue injury, and (c) to eluci- Abbreviation used in this paper: ETOH, 98% ethyl alcohol solution by the American Gastroenterological Association /89/$3.50

2 May 1989 CONTROLLED STUDY OF SCLEROSING AGENTS 1275 date the possible mechanisms of arterial coagulation and tissue injury of different sclerosants. Materials and Methods Sclerosing Agents The solutions injected were (a) 98% ethyl alcohol (ETOH), (b) a mixture with final concentration of 1% tetradecyl sulfate, 32% ethanol, and 0.3 normal saline (TES), (c) 1% polidocanol, (d) 1:10,000 epinephrine, (e) 7.2% hypertonic saline, (f) 3.6% hypertonic saline, and (g) normal saline (control). mucosa-. / needle /artery Canine Model Five healthy fasted adult mongrel dogs were studied. Anesthesia was ml of intravenous sodium thiamylal followed by endotracheal intubation and inhaled ethrane. Ethrane was administered with LI min of oxygen via a Pittman-Moore model 990 anesthesia machine. Laparotomy was performed with standard surgical techniques and blood loss was replaced with intravenous normal saline. The vital s,igns and general condition of the dog were monitored throughout the experiment by a research technician with the aid of a cardiac monitor. The small arteries supplying the canine upper small bowel were used to test the efficacy of sclerosing agents for coagulation. This model approximates the clinical treatment of arteries associated with bleeding peptic ulcers. The outside diameter of arteries studied was about 0.5 mm, which is similar to the diameter (mean, 0.7 mm) of arteries found in the base of resected gastric ulcers that have bled (23,24). After laparotomy, the small bowel was examined from its proximal to distal extent to assure normal anatomy. The small arteries penetrating the serosa of the small bowel could be distinguished from veins by arterial pulsations, red color, and thicker vessel walls. Several jejunostomies were performed beginning in the proximal jejunum and progressing distally. Jejunotomies were made on the antimesenteric surface parallel to the longitudinal axis of the bowel. Each jejunotomy was about 15 cm in length and was supplied by about eight small arteries penetrating the serosa. The jejunotomy exposed the mucosal surface of bowel supplied by these arteries and allowed observation of changes in mucosa after sclerosants were injected. Pedicles of small bowel cm in length were left undisturbed between jejunotomy sites to preserve blood flow. Randomization of Sclerosing Agents The order of sclerosing agent injection was determined by Latin square matrix with the condition that an injection of a sclerosing agent was never followed by the same agent on the subsequent round. This was to avoid possible additive effects of sclerosing agents. A research technician prepared the sclerosing agents in coded syringes and managed the order of injections. Five to six Figure 1. Schematic demonstrating injection of a serosal artery from the mucosal to the serosal side of the small bowel. The needle is directed in and around the artery as it penetrates the serosa. different arteries were injected with each solution per dog. To eliminate bias, observers were blinded to the identity of solutions until experiments and analysis of statistical data were completed. In addition, detailed descriptive data that were recorded during the experiments were correlated with the solutions injected only after completion of statistical analysis. Experiments Sequential canine gut arteries were injected with different solutions from the proximal to distal extent of each jejunotomy. Injections were performed from the mucosal to the serosal side in and around the artery in the subserosal space until arterial coagulation was achieved or a maximum volume of 3 ml was injected (Figure 1). Solutions were injected with a 26-gauge needle in increments of ml. Arteries and the adjacent bowel were examined at 1 and 5 min after the injection of each solution. Arterial coagulation and tissue injury were graded on a 0-4 scale with 0 designating no effect and 4 the greatest effect. The different effects were also reevaluated after 1 h. The grading of arterial coagulation was as follows: grade 0, no change; grade 1, absence of arterial pulsations: grade 2, decreased arterial flow; grade 3, presence of dark thrombus within arterial lumen (Figure 2); and grade 4, white coagulum involving artery and no blood flow after transection with a scalpel (Figure 3). The mucosal tissue injury resulting from injection of the different agents was as follows: grade 0, no change; grade 1, edema; grade 2, hematoma; grade 3, black discoloration (Figure 4, small arrows); and grade 4, white coagulum (Figure 4, large arrows). The serosal injury was graded similarly: grade 0, no change; grade 1, edema; grade 2, hematoma; grade 3, black discoloration; and grade 4, white coagulum. The grade was increased by one for extremely large and rapidly appearing signs of tissue injury.

3 O ~~."q., RANDALL ET AL. GASTROENTEROLOGY Vol. 96. No.5. Part " "".'--,';;" :-...",, -tl ~ N.. -'... ~ ft4' "",:.. ~~,,~ ~ ~~-..,' ; 'j ',", - ;",~.,,,,,.;,,f-'~,:,;,-~:.',;: ~:.:: -:!::', ',. ') Figure 2. Photograph of the serosa of canine gut following injection of sclerosant, showing grade 3 coagulation of a small artery with dark thrombosis within the arterial IUlllen (large arrow) and serosal injury with a blackened rim of ischemia surrounding the artery. Small arrows indicate the interruption of arterial flow by a white coagulum. Statistical Methods The grades for arterial coagulation, mucosal injury, and serosal injury of each solution were tabulated and expressed as means. The standard de'viations and standard errors were calculated and analysis of variance was performed. The paired Student's t-test was used to determine statistically significant differences between sclerosing agents and normal saline control for arterial coagulation, serosal injury, and mucosal injury. A p value of <0.05 was considered to be a statistically significant difference. Results The results for efficacy of arterial coagulation and for degree of tissue injury for each sclerosing ~,.' ""..,.1. 1'lt,~ If Figure 3. Photograph of the serosa of canine gut following injection of sclerosant showing grade 4 coagulation of the artery with white coagulum (arrows) involving the serosal artery. Figure 4, Photograph of the mucosa of the canine gut following injection of sclerosant showing grade 4 mucosal injury with white coagulum (large arrows) and a peripheral blackened zone of ischemia (small arrows). agent are shown in Figures 5 and 6, The solutions are shown in descending order of coagulation efficacy from ETOH to normal saline. The total number of arteries injected with each sclerosing agent (n) was The range in the n value is a result of randomization complexity when comparing this number of agents. The most effective agent for coagulating canine gut arteries was ETOH. Figure 3 shows the typical appearance of permanent arterial coagulation observed with ETOH. No bleeding resulted after cutting such arteries with a scalpel. Ethanol also induced serosal and mucosal injury that was significantly greater than that of normal saline control. The coagulum involving the mucosa in Figure 4 is a typical example of ETOH-induced grade 4 mucosal damage. Grade 4 serosal injury induced by ETOH also resulted in a white coagulum that involved a large area of the surrounding serosa. With ETOH, or any of the agents studied, the degree of tissue injury observed at 5 min did not change by 1 h after injection. The second most effective sclerosing agent for arterial coagulation was TES. It produced the same degree of mucosal and slightly greater serosal tissue injury than ETOH. Polidocanol coagulated arteries significantly better than normal saline control but was arithmetically less effective than ETOH and TES. The degree of mucosal injury associated with polidocanol was significant and similar to that of ETOH and TES. However, the serosal injury was not significantly different from control. Although epinephrine induced severe mucosal injury similar to polidocanol, arterial coagulation and serosal injury were not significantly different from control. Hypertonic saline solutions were not statistically different from normal saline for arterial coagulation or tissue injury.

4 May 1989 CONTROLLED STUDY OF SCLEROSING AGENTS 1277 COAGULA TION 4 r- -c 0 3 f- tel ;, C) tel 0 0 >- GI 2 r -... c( -0 GI GI... C) GI C 1 o 7.2% ri 3.6% N ETOH TES POll EPI p < 0.05 vs analysis of variance Figure 5. Histogram showing mean grade of arterial coagulation for each solution injected. Solutions are in descending order of coagulation efficacy from left to right. Agents: ETOH is 98% ethyl alcohol; TES is a mixture with final concentration of 1% tetradecyl sulfate. 32% ethanol, and 0.3 normal saline; POLl is 1 % polidocanol; EPI is 1:10,000 epinephrine; 7.2% NaCl is 7.2% hypertonic saline; 3.6% NaCl is 3.6% hypertonic saline; and N NaCl is normal saline control. Asterisks () mark solutions with coagulation efficacy significantly greater than normal saline control. Brackets are standard errors. The agents more effective than saline for arterial coagulation in decreasing order of efficacy were ETOH, TES, and POLL The correlation of experimental observations with different sclerosing agents suggested four possible mechanisms of arterial coagulation and tissue injury. Refer to Table 1. Ethanol and TES usually caused permanent arterial coagulation manifested as a white coagulum involving the artery and immediate surrounding tissue (Figure 3), a dark thrombus within the arterial lumen (Figure 2), and the absence of blood flow after artery transection. The degree of mucosal and serosal injury for ETOH and TES paralleled the degree of arterial coagulation. The effects of ETOH and TES occurred more quickly than those of other agents and the amount of tissue injury was greater when larger volumes were injected. Reduced or absent arterial blood flow was observed immediately after injection with polidocanol, with return of arterial flow by the 5 min examination. This temporary decrease in arterial flow occurred with marked bowel wall spasm following polidocanol injection and lasted for 1-5 min. Although permanent mucosal damage and edema often resulted from injection with polidocanol, arterial coagulation or serosal injury did not always follow. In contrast to ETOH and TES, the degree of mucosal tissue injury with polidocanol seemed disproportionately greater than the degree of arterial coagulation. Epinephrine frequently induced a transient compression of blood flow in serosal arteries lasting <1 min. This compression effect was also seen with saline solutions. After diffusion of the agent from the subserosal space into surrounding tissue, the arterial flow returned to normal. Arterial thrombosis or coagulation therefore did not result. As injection with epinephrine and saline solutions did not often induce arterial coagulation, the maximum 3-ml volumes were usually injected. One minute after epinephrine injection a small blanched area of mucosa (1 cm 2 ) appeared at the injection site which by 5 min progressed to a large (3 cm 2 ) zone of blackened mucosa. Observations at 1 h were similar.

5 1278 RANDALL ET AL. GASTROENTEROLOGY Vol. 96, No.5, Part 1 4 INJURY D mucosa.- c > ::I ~ serosa -0 CD "... IV <!J 2 o 7.2% 3.6% N ETOH TES POll EPI p < 0.05 vs analysis of variance Figure 6. Histogram showing mean grade of mucosal and serosal tissue injury for each solution injected. From left to right solutions are arranged in descending order of severity of induced tissue injury. See legend of Figure 5 for a definition of the solutions. An asterisk () marks an agent with a degree of tissue injury that is significantly greater than normal saline control. Brackets are standard errors. Mucosal injury (cleaf bafs) induced by ETOH, TES, polidocanol (POLl), and EPI was significantly greater than normal saline. Serosal injury (hatched bafs) induced by ETOH and TES were significantly greatef than saline control. Discussion Recently, sclerotherapy of nonvariceal GI bleeding has become popular partly because of the low cost and ease of use. Also, the agents and techniques of sclerotherapy for esophageal varices have been extensively studied (25-33) and are widely used by many GI endoscopists. Some of the agents reported to be efficacious for control of bleeding from peptic ulcers are absolute alcohol (7-10,18,21), hypertonic saline and epinephrine mixtures (12,14,15,17,18,20), sequential injection of epinephrine and polidocanol (11,16,20), and a thrombin and saline solution (13). The first reports of endoscopic injection of sclerosing agents for nonvariceal applications came from Japan. In 1975, Otani et al. (19) reported the use of ETOH for the ablation of gastric polyps and small carcinomas. In 1983, the first animal studies in this area were reported by Asaki et al. on the effect of ethanol injection in the canine gastric mucosa. Asaki et al. induced gastric bleeding by endoscopic biopsy in dogs and injected different concentrations of ETOH to assess tissue injury and hemostatic effects. The most prompt and effective hemostasis was observed with full-strength absolute alcohol. Histology revealed in situ fixation of tissue as the mechanism of hemostasis. Although a concentration of 70% ethanol was required for prompt and complete in situ fixation of tissue. concentrations of >20% had some fixative effect. Asaki was one of the first to use ethanol clinically for endoscopic control of bleeding lesions. He recommended injection of 98% ethanol in increments of ml around the bleeding site until hemostasis occurred or a maximum volume of 0.8 ml was injected (7). Another animal study of sclerotherapy for nonvariceal GI bleeding was reported by Rutgeerts et al. in 1986 (34). Bleeding ulcers were induced in the gastric mucosa of dogs by a standard ulcer maker. Injection of three different sclerosants was compared with bipolar electrocoagulation and yttrium-aluminum-garnet laser coagulation. The three sclerosing agents evaluated were absolute alcohol, 1% polidocanol, and 0.01 % epinephrine. The three sclerosing

6 May 1989 CONTROLLED STUDY OF SCLEROSING AGENTS 1279 Table 1. Proposed Mechanisms of Action and Effect of Sclerosing Agents for Arterial Hemostasis Mechanism Chemical In situ fixation Physical Arterial compression Pharmacologic Muscle spasm Effect Artery and tissue coagulation and injury. Volume dependent. Compression of artery by adjacent solution with temporary arterial flow reduction and resultant mucosal injury. Volume dependent. Agents ETOH, TES All solutions Muscle and artery Polidocanol spasm with temporary arterial flow reduction and resultant mucosal injury. Volume dependent. Arteriolar constriction Temporary Polidocanol. reduction in submucosal arteriolar flow with resultant mucosal injury. epinephrine Based on experimental observations, the different mechanisms of action and effect of sclerosing agents for arterial hemostasis gre tabulated. ETOH is a 98% ethyl alcohol solution. TES is a mixture with final concentration of 1% tetradecyl sulfate. 32% ethanol. and 0.3 normal saline. agents were injected in different volumes to assess efficacy 0f hemostasis and the resultant histologic changes. Hemostasis with alcohol injection was successful and resulted in histologic thrombosis in submucosal vessels. Tissue injury, particularly with ETOH, was a volume-dependent phenomenon. Histologic muscle necrosis and serositis were induced with injected volumes of ~1.0 ml. Tissue in the immediate zone of injection was coagulated and fixed without an inflammatory reaction. Although epinephrine controlled mucosal oozing from standard ulcers, significant bleeding could not be arrested. Histologically, epinephrine injection was associated with severe focal mucosal damage but no thrombi were found in the submucosal vessels. The only other reported laboratory data comparing the efficacy or safety of different sclerosing agents for coagulation are from our laboratory in abdominal veins or canine esophageal varices (27). We found ethanol was the most effective coagulating agent of veins but resulted in severe damage to the esopha- geal mucosa when injected endoscopically into canine esophageal varices. Injection of TES was somewhat less effective for coagulation but caused significantly less ulceration in endoscopic studies, and was the agent ultimately selected, rather than ethanol, for sclerotherapy of varices in a controlled clinical trial (26). The favorable clinical results of endoscopic sclerotherapy for nonvariceal upper GI hemorrhage for the most part have been neither controlled nor randomized (7-22). Only a few clinical investigators have evaluated sclerotherapy for hemostasis of active arterial bleeding from ulcers or for prevention of rebleeding from nonbleeding visible vessels. There is considerable variation in the techniques after initial injection as well as the choice of sclerosants. Some sclerotherapists recommend serial endoscopies with therapeutic or prophylactic reinjection of active bleeding or nonbleeding visible vessels (12-18). Some use different agents for different lesions, such as epinephrine for acute rapid bleeding versus polidocanol or ETOH for oozing or nonbleeding visible vessels (11,12,14,16). It is difficult to determine the most effective and safe sclerosing agent for control of severe ulcer bleeding from the existing literature because of the wide variation in results from different investigators and the lack of comparative or controlled data. Our laboratory studies were performed to enhance the knowledge about sclerosing agents and allow an unbiased comparison of different agents for arterial coagulation. Ethanol was clearly the most efficacious sclerosing agent for coagulation of canine gut arteries (Figures 5 and 6). The tissue injury with ETOH was similar to TES and only slightly greater than that of polidocanol and epinephrine (Figure 5). The general pattern of injury was vascular coagulation and thrombosis with serosal whitening ~r coagulation of the tissue surrounding the artery ( igures 2 and 3). A similar injury pattern was obse ved from the mucosal side (Figure 4). Our descriptive data suggested that the injection of ethanol in volumes greater than 2 ml caused severe and extensive tissue injury. Injection of <1 ml induced limited adjacent tissue injury and was still associated with arterial coagulation. The primary mechanism of action of ethanol suggested by our observations was in situ tissue fixation, which caused arterial coagulation and tissue injury. Compression of the serosal artery was a secondary effect. These mechanisms are supported by histologic work of Asaki (7) and Rutgeerts (34). Refer to Table 1. The second most effective agent for coagulation of canine gut arteries was TES. It caused the same degree of mucosal injury as ETOH but had a slightly higher grade of serosal injury. The mucosal and

7 1280 RANDALL ET AL. GASTROENTEROLOGY Vol. 96, No.5. Part 1 serosal damage patterns were similar to that of ETOH. The patterns of arterial coagulation and tissue injury suggest that TES and ETOH have similar mechanisms of action. With both, the degree of tissue damage was dependent on the total volume injected. The ability of polidocanol to coagulate arteries was inferior to that of ETOH and TES although greater than saline control. Although it often induced a temporary or partial decrease in arterial blood flow as a result of marked bowel wall spasm and edema, blood flow usually returned within 1-5 min after resolution of spasm. The most significant effect with polidocanol was the mucosal injury that resembled the effect of epinephrine. The primary mechanism of initial hemostasis seen with polidocanol was bowel wall spasm. At times this resulted in secondary arterial thrombosis and coagulation. Compression of the serosal artery was a secondary effect. Polidocanol's action on the mucosa may result from microvascular constriction secondary to muscular contraction of the gut. The histologic changes reported by Rutgeerts following injection with polidocanol were marked inflammation, tissue edema, and thrombosis in submucosal arteries (34). We did not find any significant coagulation effect on the D.5-mm arteries with epinephrine injection. Temporary compression of the serosal artery was the primary effect of the epinephrine solution. The epinephrine solution decreased arterial flow by compression, although it usually lasted for <5 min. In contrast to the effect on arteries, 1 min after injection with epinephrine there was usually a small blanched area of mucosa which progressed by 5 min to a large blackened zone consistent with mucosal ischemia. This was seen even though the penetrating serosal artery was patent and there was no significant serosal injury. The mechanism of mucosal injury with epinephrine was most likely intense constriction of the microvasculature without thrombus formation in the penetrating gut artery. This is supported by the histologic work of Rutgeerts (34) and also by in vivo microcirculation animal data reported by Guth and Smith (35). Unlike alcohol and polidocanol, Rutgeerts (34) found no thrombosis in submucosal vessels after epinephrine injection. The primary mechanism of arterial hemostasis for epinephrine and the saline solutions was compression of the artery in the subserosal space. Although these solutions diffused from the area of the serosal artery in this canine model within 1-5 min, a more prolonged compressive effect and secondary thrombosis would be predicted after injection of any solution into a closed space such as a clinical peptic ulcer. The artery within the peptic ulcer base is relatively compressible compared with the inflam- matory borders of the ulcer or the muscle layers. Transient compression of an artery with mean diameter of 0.7 mm in the ulcer base could induce secondary thrombosis and hemostasis. Conclusions The current literature about sclerotherapy for non variceal hemorrhage suggests that several sclerosing agents are efficacious for endoscopic control of ulcer bleeding. Closer scrutiny uncovers only a small body of data that deal specifically with sclerotherapy for ulcers with active bleeding or nonbleeding visible vessels. There are no controlled randomized clinical evaluations of sclerosing agents for nonvariceal bleeding similar to our present animal studies. Particularly useful for clinicians would be randomized studies of severe upper gastrointestinal bleeding that stratified patients by lesion type (such as ulcer or Mallory-Weiss) and bleeding status (active bleeding, oozing, nonbleeding clot, or nonbleeding visible vessel). Clinical mechanisms of action, efficacy, and injury patterns should be evaluated in such studies. Sclerosing agents could then be chosen based on randomized controlled clinical studies. Before the performance of such randomized clinical studies, one could make some predictions based on the results of our current experimental studies. We would predict the highest rates of permanent arterial coagulation with ETOH and a somewhat lower rate with TES, if such randomized clinical studies were performed. Ethanol and TES appear to work by two mechanisms of action: in situ tissue fixation and arterial compression. Both of these agents have the potential to cause significant tissue injury and the total volumes injected for clinical sclerotherapy of nonvariceal bleeding should be limited to reduce the extent of tissue damage. Further comparative studies in animals are also indicated to evaluate the coagulation efficacy and tissue injury of different combinations of sclerosing agents, particularly those with different mechanisms of action. References 1. Johnston JH. Jensen OM. Auth O. Experimental comparisons of endoscopic yttrium-aluminum-garnet laser, electrosurgery. and heater probe for canine gut arterial coagulation: importance of vessel compression and avoidance of tissue erosion. Gastroenterology 1987;92 : Kiefhaber p. Nath G. Moritz K. Endoscopic control of massive gastrointestinal hemorrhage by irradiation with high power neodymium-yag laser. Prog Surg 1977;15: Johnston JH. Jensen OM. Mautner W. Elashoff J. Argon laser treatment of bleeding canine gastric ulcers: limitations and guidelines for endoscopic use. Gastroenterology 1981;80: Johnston JH. Jensen OM. Mautner W. Comparison of laser and

8 May 1989 CONTROLLED STUDY OF SCLEROSING AGENTS 1281 electrocoagulation in endoscopic treatment of bleeding canine gastric ulcers. Gastroenterology 1982;82; Protell RL. Rubin CEo Auth DC. et al. The heater probe: a new endoscopic method for stopping massive gastrointestinal bleeding. Gastroenterology 1978;74: Swain CPo Bown SG. Salmon PRT. et al. Controlled trial of Nd:YAG laser photocoagulation in bleeding peptic ulcers. Lancet 1986;i: Asaki S. Tissue solidification in coping with digestive tract bleeding: hemostatic effects of local injection of 99.5% ethanol. Tohoku J Exp Med 1981;134: Asaki S. Nishimura T. Satoh A. Goto Y. Endoscopic control of gastrointestinal hemorrhage by local injection of absolute ethanol: a basic assessment of the procedure. Tohoku J Exp Med 1983;140: Sugawa C. Ikeda T. Fujita Y. Walt AJ. Endoscopic hemostasis of upper gastrointestinal bleeding by local injection of ninetyeight percent dehydrated ethanol. Surg Gynecol Obstet 1986;162: Hajiro K. Japanese experience with local injection and hemoclips. Symposium on endoscopy in upper GI bleeding. The XII International Gastroenterology Congress. The V European Gastrointestinal Endoscopy Congress. Lisbon. Portugal, September Soehendra N. Grimm H. Stenzel M. Injection of nonvariceal bleeding lesions of the upper gastrointestinal tract. Endoscopy 1985;17: Hirao M. Kobayashi T. Masuda K. et al. Endoscopic local injection of hypertonic saline-epinephrine solution to arrest hemorrhage from the upper gastrointestinal tract. Gastrointest Endosc 1985;31: Fuchs KH. Wirtz HJ. Schaube H. Elfeldt R. Initial experience with thrombin as injection agent for bleeding gastroduodenal lesions. Endoscopy 1986;18: Naka H. Matsura K. Hemostatic effect on endoscopic local injection of hypertonic saline epinephrine (HS-E) solution to arrest hemorrhage from the digestive tract. Gastrointest Endosc 1983;25: Leung JWC. Chung SCS. Endoscopic injection of adrenalin in bleeding peptic ulcers. Gastrointest Endosc 1987;332: Kortan p. Haber G. Marcon N. Endoscopic injection therapy for non-variceal bleeding lesions of the upper gastrointestinal tract (abstr). Gastrointest Endosc 1986;32: Chung SCS. Leung JWC, Steele RJC, Crofts TJ. Epinephrine injection for actively bleeding ulcers: a randomized controlled study (abstr). Gastrointest Endosc 1987;33: Chen PC. Wu CS. Liaw YF. Hemostatic effect of endoscopic local injection with hypertonic saline-epinephrine solution and pure ethanol for digestive tract bleeding. Gastrointest Endosc 1986;32: Otani T. Tatska T. Kanamura K. et al. Intramural injection of ethanol under direct vision for the treatment of protuberant lesions of the stomach. Gastroenterology 1975;69: Rutgeerts P. Broekaert L. Coremans G. et al. Randomized comparison of three hemostasis modalities for severely bleeding peptic ulcers: epinephrine 0.01% injection alone (1). epinephrine + polidocanol 1% injection (2). epinephrine injection followed by YAG laser (3) (abstr). Gastrointest Endosc 1987;33: Sato W. Komatsu K. Moriai. et al. A comparative study of endoscopic hemostasis for upper gastrointestinal bleeding with reference to Nd-YAG laser photocoagulation. Gastrointest Endosc 1984;26: Rutgeerts p, Vantrappen G. Coremans G. Broeckaert L. Janssens J. Pretreatment of bleeding ulcers with adrenalin injections renders YAG laser photocoagulation highly efficacious (abstr). Gastroenterology 1983;84: Johnston JH. The sentinel clot and invisible vessel: pathological anatomy of bleeding peptic ulcer. Gastrointest Endosc 1984;30: Swain CP, Storey DW. Bown SG. et al. Nature of the bleeding vessel in recurrently bleeding gastric ulcers. Gastroenterology 1986;90: Jensen DM. Sclerosants for injection sclerosis of esophageal varices. Gastrointest Endosc 1983;29: Jensen DM, SUpa M. Reedy T. Elashoff J. and the CURE Hemostasis Research Group. Controlled trial of sclerotherapy for bleeding esophageal varices in alcoholic cirrhosis (abstr). Gastroenterology 1986;90: Jensen DM, Machicado GA. SUpa M. Esophageal varix hemorrhage and sclerotherapy-animal studies. Endoscopy 1986; 2: Johnson GW. Ridgers HW. A review of 15 years' experience in the use of sclerotherapy in the control of acute hemorrhage from oesophageal varices. Br J Surg 1973;60: Piquet KJ. Fleig WE. Sclerotherapy of esophageal varices. In: Papp JP. ed. Endoscopic control of gastrointestinal hemorrhage. Boca Raton. Fla: CRC Press. 1981: Terblanche I. Northover JMA. Bornman P. et al. A prospective evaluation of injection sclerotherapy in the treatment of acute bleeding from esophageal varices. Surgery 1979;85: Lewis I. Chung RS, Allison J. Sclerotherapy of esophageal varices. Arch Surg 1980;115: Sivak MV. Stout PI. Skipper G. Endoscopic injection sclerosis of esophageal varices. Gastrointest Endosc 1981;27: Hughes RW. Larson DE. Biggiano TR. et al. Endoscopic variceal sclerosis: a one-year experience. Gastrointest Endosc 1982;28: Rutgeerts P, Geboes K. Vantrappen G. Tissue damage produced by hemostatic injections (abstr). Gastrointest Endosc 1986;32: Guth PH. Smith SE. Vasoactive agents and the gastric microcirculation. Microvasc Res 1974;8: Received December Accepted November Address requests for reprints to: Dennis M. Jensen. M.D. GI Division CHS, Department of Medicine. UCLA School of Medicine. Los Angeles. California This study was supported in part by Veterans Administration Research Funds. The authors thank Mary Crump for secretarial assistance and Terry Reedy for statistical assistance.