Injection therapy for nonvariceal gastrointestinal bleeding

G astrointestinal hemorrhage is a common and serious problem its average mortality of 10% has changed little over the past 40 years. In 80% of patients the bleeding stops spontaneously. In patients with continuous or recurrent bleedmg (20%), mortality and morbidity are high , and emergency surgery is often required, which has a higher mortality than the same operation performed e lectively. Succes.sful the rapeutic endoscopic mtervenuon in this h igh nsk group is necessary to improve outcome. For injection treatment of no nvariceal bleeding lesions, the autho r has been usmg the Soehendra method (l : I 0,000 adrenaline and poltdocanol) with success m 90'10 of actively bleeding patients. T h ree con t rollt·d tria ls of endoscopic sclerosis in bleeding peptic ulcer disease showed decreased blood transfusions, surgery and hospita l stay, but did not find any significant difference m mortali ty. The ideal solution and the usefulness of addit ional therapy are questions which must be addressed v ia prospective controlled trialsofa large numberof pat ients. CanJ Gastroenterol 1990;4(9):650-652 Key W ords: Adrenaline, Gastrointestinal bleeding, Hemostasis, Injection chera/JY Traitement par injection des hemorragies gastrointestinales non variqueus.es RESUME: L'hemorrag1e gastro-incestmale est un probleme courant serieux qui s'accompagne d'une morra lice moyenne de 10 %, chiffre qui a peu change au co urs des quarante dernieres annees. Chez 80 % des patients, l'hemorragie cesse spontanemen t. Chez Jes patients qui saignent consrnmment ou done les sa1gnemen ts recidivent (20 %), la morta lite e t la morbid ite sont elevces et une intervention d 'urgence s'impose souvent, laquelle est assort1e <l'un taux de morta lite supcrieure a celle des memes operations a froid. Dans cc groupe a risque e leve, ii est n ccessaire de recou ri r aux interve n t ions en doscopiq ues therapeuttqucs afin d'ameliorcr les resultats. L'auteur a recours au t ra1tement par injection des lesions non variqueuses qui saignent; ii ut ilise la methode de Soehendra (1 : IO 000 adrenaline et polidocanol) avec succes chez 90 % des patients qui sa ignent acttvemcnt. Trois essa is controles units.mt des injections sclerosantes effectuees sous endoscopic dans le traitement des ulceres gastro duodenaux hemorragiq ues rapporten t une diminution des transfusions, des gestes D1111swn of Gastroenterolo,a, University of Tcmmw, T he WI ellesley / losp,ral , Toronw, Onrano Correspondence and reprmrs. Dr P Korum , Assrsranr Professm, Dit•1s1cm of GastroenwroloJ:Y, University of Toronto, The Wlellesle"I I losp1tal, 160 Wlellesley Street Ea..11, Swte 41 9, Jones Building, Toronw, Onrano M4Y I) 3 GASTROINTESTINAL I IEMORRHAGE 1s a common and serious problem with an average mortality of I 0% (I). It is often assumed that early correct Jrag· nosis will re<luce mortality rate,; however, several control led studies hal'e shown no such advantage, anJ there has been little change in morrnlicyover the past 40 years. S ince the va11 maiorr t y o f uppe r gas trointcmnal bleeding episodes arc sclf-lim1ceJ, it ha, been c~timated tha t more than SOOJ patients would have to be ranJorn1:eJ before a tria l could demonstrate that endoscopy decreases mortality. More impo rtan tly, the maiority of endoscopic tria ls were carried our before the cvolutton of therapeutic technique~ that cm po tentia lly con t rol gastrointc~tmal bleeding (2,3 ). In 80% of patients, the hlecdmg stops spo ntaneously and the hospital cour e 1s usually benign. However, m patients with continuous or recurrent bleeding ( 20%), mortality and rnorb1d1ty are high and emergency surgcr, is often needed ( 4) . Emergency surgery for upper gastrointest inal bleeding isa1sociate<l with a higher mortality than for the same operauon performed elec, t 1vely. S uccessfu l the rapeutic endoscopic intervention in thb high ris~ group 1s necessary to 11nprove the out· come. A numbe r o f endoscopic therapeutic m0Jalit1e~ are now avail, able to t reat upper gastromtesnnal 650 CAN J UA~TROENTEROL VOL 4 No 9 DECEMBER 1990 chirurgicaux et des hospitalisations, mais ne montrent aucune difference significative quanc a la mortalite. ll faudrait examiner quelles seraient la solution idfale et l'utilite d'un traitemcnc additionnel a l'aide d'essais prospectifs controles portant sur un grand nombre de patients. bleeding. These include yttriumaluminum-garnet (Y AG) laser, monopolar and bipolar coagulation, heater probe and injection therapy. None of these modalities has been sufficiently well studied to permit definitive recommendations. The ideal hemostatic technique should be simple, safe, portable, inexpensive and effective. These criteria are best fulfilled by injection therapy. Injection sclerotherapy has hcen used extensively in the control of bleeding esophageal vances and over the past few years has gained popularity for nonvariceal bleeding. The technique has been practised in Japan and West Germany since the mid 1970s (5). A number of different substances have been injected, most of which are vasoconstrictors or sclcrosancs. Table 1 lists the injectable agents that have been u~ for treatment of upper gastrointestinal bleeding. TECHNIQUE Once an active or potential bleeding site has been accurately located, injection therapy can be attempted. The injection is performed using a variceal injecwr. A large single or double channel endoscope is preferred to allow for adequate suctionin g of blood . Sclerosant or vasoconstrictor is injected at three or four sites surrounding an exposed bleeding vessel or nonbleeding visible vessel. The volume of agent injected depends on the type and should be minimized to avoid extension of ulcer and transmural injury. MECHANISMS OF ACTION The hemostatic mechanism of injectable agents is based on vasoconstricting or sclerosing effect. Adrenaline in animal experiments has been shown co cause a prolonged decrease in local gastric bloodflow (68). The decrease in flow is felt co be secondary to the vasoconstrictive effect of adrenaline rather than to a mechanical effect of the injected volume. Injection of absolute ethanol has been shown to cause degeneration and necrosis of the blood vessel wall with consequent thrombosis (9,10). This is accomplished by the fixative effect of ethanol. Part of the hemostasis may be supported by mechanical compression of vessels. Hypcrtonic saline was shown to cause tissue edema, degeneration of vascular wall and consequent thrombosis in the vascular lumen ( I l). Rutgeerts (12) examined the tissue effects of adrenaline 1: 10,000, absolute alcohol and 1 % polidocanol, anti compared them with Y AG laser and BICAP in dogs. Adrenaline caused focal damage without thrombosis of submucosal vessels and without serositis. Ethanol caused dehydration and fixation of the tissue with extensive suhmucosal and muscular necrosis and vessel thrombosis. Serositis was frequently observed. There was a linear relationship between the extent of damage and the volume of solution injected. Injection of I% polidocanol cause<l lesions characterized by mucosa I necrosis, partial necrosis of submucosa and submucosa l edema. There was partial thrombosis of the vessels. The extent of damage from l % polidocanol was related to the volume injected. CLINICAL EXPERIENCE Uncontrolled studies have been reported on the use of injection treatment with a number of <lifferent hemostatic agents, with successful treatment ranging from 84 to 100%. Asaki (9) injected absolute alcohol and obtained hemostasis in 10096 of patients with I 0% reblee<ling. Sugawa ( 13) also using alcohol obtained permanent hemostasis in 88% of patients. Soehendra ( 14) has been injecting adrenaline 1: 10,000 combined with l % polidocanol and obtained hemostasis m 84% of pulsatile CAN J GASTROENTEROL VOL 4 No 9 DECEMBER 1990 Injection therapy for nonvariceal GI bleeding TABLE 1 Injectable agents used for treatment of upper gastrointestinal bleeding Adrenaline l: 10,000 Adrenaline and hypertonic saline solution Adrenaline l : 10,000 followed by polldocanol Absolute ethanol Thrombln Ethanolamine bleeders. Leung ( 15) used adrenaline alone with permanent hemostasis in 92% of patients. Rutgeerts anc..l colleagues ( 16) randomized 140 patients with ulcers cont a in ing an actively hleeding or nonhleeding visible vessel to sham treatment or to one of three endo~copic methods of hemostasis: adrenaline l:10,000 alone, adrenaline 1:10,000 plus I 'Xi polidocanol, and adrenaline followed by YAG laser treatment. In patients with a nonbleeding visible vessel. the sham treatment was significantly less effective than adrenal ine plus polidocanol or adrenaline plus laser in achieving hemostasis. All three treatmen rs reduced total transfusion requirement sign ificantly compared with sham treatment. Adrenaline plus polidocanol was significantly more effective than adrenaline alone in achieving permanent hemosrnsis. Adrenaline plus laser was also more effective than adrenaline alone, but this difference was not statistically significant. The authors concluded that injection therapy with adrenaline plus poli<locanol should be the treatment of choice hascd on cost, ease of use and safety. A recent report from Spain documents the efficacy of this technique in a controlled trial (l 7). One hundred an<l thirteen patients with hemorrhage were randomized to endoscopic injection therapy and Hz blockers, or to 112 blockers alone. Each group had an equal number of patients with hleeding and nonhlee<lmg visible vessels. All les,ons were injected with 3 rn 10 mL of adrenaline I : 10,000 and 6 to 12 mL of I% polidocanol. Major recurrent hlceding occurre<l in 5% of injected patients :md in 23% of patients randomized to

G ASTROINTESTINAL I IEMORRHAGE 1s a common and serious problem with a n average mortality of I 0% (I).It is often assumed t hat early correct Jrag• nosis will re<luce mortality rate,; however, several control led studies hal'e shown no such advantage, anJ there has been little ch ange in morrnlicyover the past 40 ye ars.S ince th e va11 maiorr t y o f uppe r gas trointcmnal bleeding episodes arc sclf-lim1ceJ, it ha, been c~timated tha t more than SOOJ patie nts would have to be ranJorn1:eJ befo re a tria l could demonstrate that endoscopy decreases mortality.More impo rtan tly, the maiority of endoscopic tria ls were carried our before the cvolutton of therapeutic technique~ that cm po tentia lly con t rol gastrointc~tmal bleeding (2,3 ).
In 80% of pati ents, the hlecdmg stops spo ntaneously and the hospital cour e 1s usually benign .However, m patients with continuous or recurrent bleeding ( 20%), mortality and rnor-b1d1ty are high and emergency surgcr, is ofte n needed ( 4) .Emergency surgery for upper gastrointestinal bleeding isa1-sociate<l with a higher mortality than for t he same opera uon performed elec, t 1vely.S uccessfu l the rapeutic endoscopic intervention in thb high ris~ group 1s necessa ry to 11nprove the out• com e.A numbe r o f e ndoscopic therapeutic m0Jalit1e~ are now avail, able to t reat uppe r gastro mtesnnal chirurgicaux et des hospitalisations, mais ne montrent aucune difference significative quanc a la mortalite.ll faudrait examiner quelles seraient la solution idfale et l'utilite d'un traitemcnc additionnel a l'aide d'essais prospectifs controles portant sur un grand nombre de patients.
bleeding.These include yttriumaluminum-garnet (Y AG) laser, monopolar and bipolar coagulation, heater probe and injection therapy.None of these modalities has been sufficientl y well studied to permit definitive recommendations.
The ideal hemostatic technique should be simple, safe, portable, inexpensive and effective.These criteria are best fulfilled by injection therapy.Injection sclerotherapy has hcen used extensively in the control of bleeding esophageal vances and over the past few years has ga ined popularity for nonvariceal bleeding.The technique has been practised in Japan and West Germany since the mid 1970s (5).A number of different substances have been injected, most of which are vasoconstrictors or sclcrosancs.Table 1 lists the injectable agents that have been u~ for treatment of upper gastrointestinal bleeding.

TECHNIQUE
Once an active or potential bleeding site has been accurately located, injection therapy can be attempted.The injection is performed using a variceal injecwr.A large single or double channel endoscope is preferred to allow for adequate suctionin g of blood .Sclerosant or vasoconstrictor is injected at three or four sites surrounding an exposed bleeding vessel or nonbleeding visible vessel.The volume of agent injected depends on the type and should be minimized to avoid extension of ulcer and transmural injury.

MECHANISMS OF ACTION
The hemostatic mechanism of injectable agents is based on vasoconstricting or sclerosing effect.
Adrenaline in animal experiments has been shown co ca use a prolonged decrease in local gastric bloodflow (6)(7)(8).The decrease in flow is felt co be secondary to the vasoconstrictive effect of adrenaline rather than to a mechanical effect of the injected volume.Injection of absolute ethanol has been shown to cause degeneration and necrosis of the blood vessel wall with consequent thrombosis (9,10).This is accomplished by the fixative effect of ethanol.Part of the hemostasis may be supported by mechanical compression of vessels.
Hypcrtonic saline was shown to cause tissue edema, degeneration of vascular wall and consequent thrombosis in the vascular lumen ( I l).

CLINICAL EXPERIENCE
Uncontrolled studies have been reported on the use of injection treatment with a number of <lifferent hemostatic agents, with successful treatment ranging from 84 to 100%.Asaki (9) injected absolute alcohol and obtained hemostasis in 10096 of patients with I 0% reblee<ling.Sugawa ( 13) also using alcohol obtained permanent hemostasis in 88% of patients.Soehendra ( 14)

TABLE 1 Injectable agents used for treatment of upper gastrointestinal bleeding
Adrenaline l: 10,000 Adrenaline and hypertonic saline solution Adrenaline l : 10,000 followed by polldocanol Absolute ethanol Thrombln Ethanolamine bleeders.Leung ( 15) used adrenaline alone with permanent hemostasis in 92% of patients.
Rutgeerts anc..l colleagues ( 16) randomized 140 patients with ulcers cont a in ing an actively hleeding or nonhleeding visible vessel to sham treatment or to one of three endo~copic methods of hemostasis: adrenaline l:10,000 alone, ad renaline 1:10,000 plus I ' Xi polidocanol, and adrena line followed by YAG laser treatment.In patients with a nonbleeding visible vessel.the sham treatment was significantly less effective than adrenaline plus polidocanol or adrenaline plus laser in achieving hemostasis.All three treatmen rs reduced total transfusion requirement sign ifi cantly compared with sham treatment.Adrenaline plus polidocanol was significantly more effective than adrenaline alone in achieving permanent hemosrnsis.Adrenaline plus laser was also more effective than adrenaline alone, but this difference was not statistically significant.The authors concluded that injection therapy with adrenaline plus poli<locanol sho uld be the treatment of choice hascd on cost, ease of use and safety.
A recent report from Spain documents the efficacy of this technique in a controll ed trial (l 7).One hundred an<l thirteen patients with hemorrhage were randomized to endoscopic injection therapy and Hz blockers, or to 112 blockers alone.Each group had an equal number of patients with hleeding and nonhlee<lmg visible vessels.All les,ons were injected with 3 rn 10 mL of adrenaline I : 10,000 and 6 to 12 mL of I% polidocanol.Major recurrent hlceding occurre<l in 5% of injected patients :md in 23% of patients randomized to treatment wah I lz blockers alone (P<0.001 ).The neeJ for emergency surgery was reduccJ from 34 to 5% ( P<0.00 I) and there was n significantly reduccJ requirement for blood transfusions and a reduction in the length of hospital stay.
A scconJ randomized stutly by Chung et al ( 18) compared iniection of aJrenaline into blcetling ulcers with no endoscopic treatment.There was a significant difference in the two groups with respect to requirement for emergency surgery, requirement of bkxxl transfusion anti median hospital stay.
The th1rJ prmpect1ve randomized trial was carried out by Balanzo ct al (l 9).They stutlied 72 patients with bleeding ulcers.Half were rantlomized to inJection treatment with adrenaline plus poli<locanol and half were managed medically.The study again confirmed the benefit of inject10n therapy comparing the requirement for emergency surgery and blooJ transfusions.
None of these studies was able to show a significant difference m mortality.

COMPLICATIONS
The use of polidocanol has been as, soc.iatcJ with the appearance of large ulcers when more than 20 mL have been iniccted.The perforation rate is l % (20).ln1cction ot absolute alcohol has been associated with a less than I ' X, has been injecting adrenaline 1: 10,000 combined with l % polidocanol and obtained hemostasis m 84% of pulsatile CAN J GASTROENTEROL VOL 4 No 9 DECEMBER 1990 Injection therapy for nonvariceal GI bleeding Gehocs K, Vantrappcn G.