- Email: [email protected]
A controlled comparison of continuous intraarterial and intravenous infusions of vasopressin in hemorrhage from esophageal varices
GASTROENTEROLOGY
77:540-546,1979
A Controlled Comparison of Continuous Intraarterial and Intravenous Infusions of Vasopressin in Hemorrhage from Esophageal Varices M. CHOJKIER, S. BAR-MEIR, J. L. KNIAZ,
R. J. GROSZMANN, C. E. ATTERBURY, A. T. BLEI, J. FRANKEL, M. G. GLICKMAN, R. SCHADE,
G. J. TAGGART,
and H. 0.
CONN
Liver Research Laboratory, Veterans Administration Medical Center, West Haven, Connecticut, and The Department of Internal Medicine, Yale University School of Medicine, New Haven, Connecticut
Infusions of intraarterial vasopressin (IAV) into the superior mesenteric artery have been shown to be effective in controlling hemorrhage from esophagogastric varices. Intravenous infusions of vasopressin (NV), which can be initiated rapidly and require less sophisticated equipment and personnel, have also been reported to control variceal hemorrhage. We undertook a controlled clinical trial to compare these two routes of administration. Twenty-two cirrhotic patients with massive hemorrhage from varices were randomized to receive either IVV or IAV. Intraarterial vasopressin was begun at 0.2 U/min and increased progressively as needed to 0.2,0.3, 0.4, and 0.5 U/min. Intravenous vasopressin was begun at 0.3 U/min and increased progressively as needed to 0.6,0.9,1.2,and 1.5 U/min. Hemorrhage was controlled in 5 of 10 episodes (50%) with IVV and in 6 of 1.2episodes (50%) with IAV. Seven of the ten episodes treated with IVV (70%) ended fatally compared with 9 of 12 treated with IAV (75%). Side-effects and complications occurred with similar frequency in the two groups. The two routes of administration are equal in effects, side-effects, and complications. We recommend that NV, which can be administered more easily, be given a brief therapeutic trial early in the management of hemorrhage from varices.
Posterior substance,
pituitary extract, a potent vasoconstrictive was reported in 1956 to be useful in the
Received January 3,1979. Accepted May 11,1979. Address requests for reprints to: H. 0. Conn, M.D., VA Medical Center, West Haven, Connecticut 06516. 0 1979 by the American Gastroenterological Association 0016-5085/79/090540-07$02.00
treatment of bleeding esophageal varices.’ During the succeeding 20 yr, controlled clinical trials have shown that posterior pituitary extract, or its vasoactive principle vasopressin, is effective in controlling hemorrhage from esophageal varices. First, Merigan et al. showed that intravenous, bolus injections were successful.’ Subsequently, two groups found that continuous infusions of vasopressin into the superior mesenteric artery controlled variceal bleeding.“,’ Most recently, Johnson et al. reported that continuous intravenous vasopressin was as effective as intraarterial infusions in the treatment of hemorrhage from varices.’ Several of these report? and others’-* suggested that vasoconstrictive therapy might be effective in nonvariceal bleeding lesions as well. Continuous infusions of vasopressin into the superior mesenteric artery had been introduced to decrease portal pressure while avoiding the systemic effects observed with the bolus intravenous injection of vasopressin.“-” Recent studies have suggested that the hemodynamic responses, complications, and the therapeutic effects of intravenous vasopressin infusions”.‘2-‘5 were similar to those of intraarterial infusions.“-G.‘“~ZG Indeed, Merigan and his colleagues at the Boston City Hospital showed in the first prospective randomized investigation of vasoconstrictive therapy that although the bleeding was controlled, the mortality rate was not significantly decreased.’ Unfortunately, this early experience appeared to set the pattern for subsequent studies.“-’ The simplicity, speed, and apparent success of the systemic intravenous administration of vasopressin, which requires neither experienced angiographers nor expensive, sophisticated equipment, suggest that this route of administration might be more practical
September 1979
INTRAARTERIAL AND INTRAVENOUS VASOPRESSIN IN GIHEMORRHAGE
than selective intraarterial infusion. The only published controlled comparison of intravenous and intraarterial vasopressin, which showed the two routes to be equally effective, was performed in a relatively small group of patients.” The present investigation was undertaken to compare the effect of selective intraarterial infusions of vasopressin (IAV) with systemic intravenous vasopressin (NV) on the control of upper gastrointestinal hemorrhage, on survival, and on the frequency, nature, and severity of the systemic complications of vasopressin therapy.
Methods and Materials Patients with massive upper gastrointestinal (GI) bleeding admitted to the West Haven Veterans Administration Hospital between May 1976 and December 1977 were candidates for inclusion in this investigation. Massive hemorrhage was defined as the loss of 1000 ml or more
blood,
confirmed
by a drop in hematocrit of at least of vital signs (pulse rate of > lOO/ min, blood pressure < 100/60mm Hg, or a decrease in blood pressure of > 20/10mm Hg on the patient’s assuming the sitting position). Patients with known coronary or cerebral artery disease were excluded. After informed consent had been obtained, patients were randomized by sealed envelope technique to receive either IVV or IAV infusions in addition to conventional medical management. The randomization scheme was devised by Mr. E. Lerner, a biostatistician, whose assistance is acknowledged, using a table of random numbers. Decisions about eligibility were made by the postdoctoral fellows of the Liver Research Laboratory and the responsible house officer and confirmed by the attending physician before randomization. Conventional medical management is defined as any form of treatment except vasoconstrictive therapy. Separate randomization schemes were used for patients bleeding from esophagogastric varices and for those bleeding from other, nonvariceal sites. This investigation was approved by the Human Investigations Committee on January 23,1976. The nature of the investigation and the potential dangers of vasopressin infusion were explained to the patients and/or their next of kin, and written authorization for participation in the investigation was obtained. The decisions about the control of hemorrhage in individual patients were made retrospectively on prospectively recorded data by two of the investigators (MC and HOC) independently. The investigators, who had no preconceived opinions and who made these decisions solely on the basis of predetermined criteria, agreed in each instance. Student’s paired and unpaired t-tests and x2 tests were used in statistical analyses. Only seven episodes of GI hemorrhage from nonvariceal sources were randomized, a number too small to evaluate, and, consequently, this report deals primarily with those patients who bled from varices. Twenty-two episodes of hemorrhage from varices in 19 patients satisfied these criteria and were randomized in this investigation. All were men who ranged in age from
6% and by instability
541
25 to 77 (mean 54) yr of age. All were cirrhotic patients except one, who was shown subsequently to have a portal vein thrombosis. Attempts were made to establish the site of hemorrhage by endoscopy and/or arteriography in virtually all. Endoscopy was performed in 20 of the 22 episodes (91%). In one patient, previous endoscopic examinations had shown esophageal varices, and arteriography was done instead. Another patient developed shock before endoscopy could be performed, and randomization was carried out without a definitive diagnosis. Arteriography (64%).In five epiwas performed in 14 of the 22 episodes sodes, the bleeding site had been identified by endoscopy, and arteriography was considered unnecessary. In one instance, the patient developed shock and died before angiography could be undertaken. In 2 patients, the bleeding stopped spontaneously, and no further studies were performed. The site of hemorrhage was considered proved if the bleeding lesion was demonstrated at endoscopy, probable, if varices were identified in the absence of any other potential bleeding lesion, and clinical if not so clearly established. Two patients who had massive variceal hemorrhage were excluded from this investigation. One was excluded because of coronary artery disease, and he subsequently died. The other was excluded because the bleeding had stopped spontaneously; he survived. The site of hemorrhage was proved in 7, probable in 11, and clinical in 4 of the 22 episodes. Ten patients were randomized to receive IVV and 12 to receive IAV. Three patients were randomized twice, each on separate hospital admissions. One of them was selected for IVV on both occasions, one for IAV twice, and the third was treated once by each route. Each patient had a nasogastric tube in place, and serial gastric aspirations were obtained. Hemorrhage was considered controlled by vasopressin if bleeding was stopped for a period of at least 24 consecutive hr, as evidenced by the absence of blood in the gastric aspirations and by stability of vital signs and hematocrit. Vasopressin was administered continuously with a Harvard pump at a concentration of 1 Unit/ml of physiologic saline. Intraarterial infusions were begun at 0.1 U/min for at least 30 min, and if ineffective, the dosage was progressively increased at 30- to 60-min intervals to 0.2, 0.3, 0.4, and 0.5 U/min. Intravenous infusions were begun at 0.3 U/min for at least 30 min and, if ineffective, were progressively increased at approximately 30- to 60-min intervals to 0.6, 0.9, 1.2, and 1.5 U/min. The duration of hemorrhage before the institution of therapy was calculated from the time the bleeding began, usually outside of the hospital, until the time vasopressin was begun. All patients were treated in an Intensive Care Unit with continuous cardiac monitoring. If treatment was unsuccessful, depending on the rate of bleeding and the clinical situation, therapy was declared a failure and the patient could thereafter be treated with the other route of vasopressin administration, which is hereafter referred to as crossover therapy. Esophageal tamponade was used in
only 1 randomized patient. The balloon tube was passed while awaiting arterial catheterization. Recurrent hemorrhage was defined as renewed bleeding
CHOJKIER ET AL.
542
Table 1.
Comparsion
GASTROENTEROLOGY
of Laboratory
Values in WV and IAV Groups
Observation No. of episodes Age (~4 Duration of hemorrhage (hr) Hematocrit at onset of therapy (%) Hemoglobin at onset of therapy (g%) Ascites (a) Encephalopathy (%) Serum albumin (g%) Blood urea nitrogen (mg%) Total serum bilirubin (mg %) Direct serum bilirubin (mg %) Serum alkaline phosphatase (ILJ) Serum aspartic aminotransferase (IU) Total serum protein (g%) Prothrombin
time prolongation
Data expressed
IAV group
Statistical significance”
10
12 52 f 7 15 f 11 29f5 10.2-+1.5 50 17 3.1+ 0.5 31f25 6.2 -t5.6 3.7+ 4.3 158 f 77 63f47 6.7 + 1.1 3.4f 2.6
NS NS
Results Comparison
of Treatment
Groups
The groups selected for IVV and for IAV were similar at the time of randomization in age, diagnosis, clinical findings, and laboratory values (Table 1). They were also similar in the severity of hemorrhage as manifested by the duration of the hemorrhage, the hematocrit and hemoglobin concentrations, and the number of transfusions administered at the time of randomization and at the time therapy started (Tables 1 and 2). The mean interval between randomization and the initiation of therapy was 45 min for IVV and 3 hr for IAV. Comparison Therapeutic
NS
NS NS NS NS NS NS NS NS NS NS NS
Control
of Hemorrhage
Bleeding was controlled with IVV in 5 of 10 episodes (50%) randomized to IVV and in 6 of 12 episodes (50%) selected for IAV (Figure 1). In 3 patients selected for IAV therapy and in 1 randomized for IVV, bleeding stopped before vasopressin was administered. In one instance, a patient randomized to receive IAV died before vasopressin therapy was begun. Inclusion or exclusion of these nonconforming patients does not alter the results appreciably. Thus, bleeding was actually controlled by the administration of IVV in 4 of 9 episodes and by IAV in 3 of 8 episodes. Nonvariceal bleeding was controlled in 1 of 3 patients treated with IVV and in 1 of 4 with IAV. Crossover therapy was employed on nine occasions. In 5 episodes, IVV failed to control the hemorrhage and IAV was tried. In 3 of the 5 (60%), therapy with IAV was successful. In 4 episodes, IAV failed to control hemorrhage and IVV was then used. In none of the 4 was IVV treatment successful. These
of WV and ZAV in the Treatment of Variceal Hemorrhage index
No. of episodes Control of hemorrhage Survival Duration of vasopressin therapy (hr) Amount of vasopressin infused (U) Transfusion prior to vasopressin (U) Transfusions with vasopressin (U) Rate of transfusions prior to vasopressin (U/hr) Rate of transfusions with vasopressin (U/hr) Major complications Minor complications Data expressed
NS
as mean + SD. n NS = Not significant.
within seven days of the successful treatment of bleeding varices. Survival was defined as the discharge of the patient, alive, from the hospital during the admission in which the bleeding occurred.
Table 2.
at Time of Randomization
IVV group 56 f 15 16 f 11 30 +3 10.3-t1.2 50 20 3.2-c0.7 32 +14 3.8-c3.7 3.2f 6.2 138r?:72 190+281 6.3 -+1.3 3.3+ 3.3
(set)
Vol. 77, No. 3
IVV
IAV
5(50%) 3(30%) 18 +ll 54O-t285 4.3 -c1.2 3.7+ 3.8 0.26 0.20 3 4
12 6 (50%) 3(25%) 16 +lO 221*135 5.4f 3.0 3.4 f 4.6 0.36 0.21 3 2
10
as Mean f SD. *P c 0.02 (dosage determined by treatment allocation) 0 NS = Not significant.
Statistical significance” NS NS NS * NS NS NS NS
NS NS
Scptembcr
INTRAARTERIAL
1979
AND INTRAVENOUS VASOPRESSIN IN Gl HEMORRHAGE
data suggest that IAV may be more effective than IVV when the “other” route of administration has failed. Recurrent hemorrhage occurred after successful therapy with IVV in 2 of 5 patients (40%). In 1 of these patients, IVV was again used and was successful. Recurrent bleeding was observed in 3 of the 6 patients (50%) controlled with IAV. In only 1 of the 3 patients was bleeding again stopped by IAV. These differences are not statistically significant.
Survival Seven of the 10 episodes (70%) treated with IVV ended fatally, compared with 9 of the 12 (75%) treated with IAV. This difference is not statistically significant (Figure 2). Two of 3 patients with nonvariceal bleeding treated with IVV died, as did all 4 who received IAV.
Duration
of Therapy
Before admission to the study, the IVV and IAV groups had been bleeding for a mean of 16 and 15 hr, respectively. In the IVV group, the duration of therapy ranged from 5 to 32 hr and averaged 18 hr. In the IAV group, the superior mesenteric artery was infused. Vasopressin was infused for periods of 3 to 27 hr, with a mean of 16 hr. The catheter remained in place, including saline infusions, for an average of 32 hr, and one remained in place for 97 hr. In 1 patient in whom two separate catheterizations were performed, the total duration of catheterization was 121 hr. More vasopressin was infused into the patients by the intravenous route (mean, 540 U) than by the intraarterial route (mean 221 U), a significant difference (P < 0.02). This difference reflects the higher concentration of vasopressin given intravenously.
Transfusion
Requirements
Before admission to the study, the IVV group received an average of 4 U of blood compared with 5 U in the IAV group. Thus, the patients in the two groups had received 0.26 and 0.36 U of blood per hour of hemorrhage, respectively (Table 2). These differences are not statistically significant. After vasopressin was begun, the IVV group received an average of 4 U of blood compared with 3 U with the IAV group (Table 2). Thus, during vasopressin therapy, the IVV group received 0.20 U/hr and the IAV group was transfused with 0.21 U/hr. Neither of these differences is significant.
543
Complications Among the IVV group, three patients had major complications. One had a cardiorespiratory arrest after 10 hr of treatment at a maximal rate of 0.9 U/min. An idioventricular rhythm of 30 per minute and extreme acidosis (pH 6.95) were corrected, but bleeding persisted, and the patient died of hemorrhage 12 hr later. A second patient had phlebitis at the site of the infusion and staphylococcal septicemia. Antibiotic therapy was successfully administered. The third complication in the IVV group may actually have been caused by IAV therapy. After IVV had been administered for 10 hr at a rate up to 1.5 U/min without controlling hemorrhage, IAV was administered for 10 additional hr, at which time the patient developed shock. The patient died 20 hr after IAV had been stopped, and autopsy showed ischemic necrosis of the large bowel. Four additional patients had minor complications such as hypertension, bradycardia, or ventricular extrasystoles. Among the IAV group, 3 patients had major complications and 1 patient developed bacteremia, with a Klebsiella species and Staphylococcus aureus. The staphylococcus was also cultured from the tip of the catheter. One patient developed acute pulmonary edema after 1 hr of successful vasopressin administration at the rate of 0.3 U/min. Vasopressin was stopped and the cardiac status improved, but when bleeding recurred, IAV was again started and pulmonary arterial wedged pressure rose alarmingly, and IAV was again discontinued. One patient bled massively (estimated blood loss of 1000 ml) from the site of a prolonged and difficult catheterization. Two other patients experienced clotted catheters, which required replacement.
Discussion This investigation supports the previous observations of Johnson et al. that continuous intravenous infusions of vasopressin are as effective in the treatment of bleeding varices as infusions of vasopressin into the superior mesenteric artery.s It also confirms numerous reports that the same systemic side-effects and complications occur with IAV as with intravenous administration.“~“,‘~-‘~,~l-~~ Unfortunately, the two controlled clinical trials together include only 47 patients, far too few to exclude the possibility that one of these routes of administration is truly superior to the other. The large number of additional patients that would be needed to eliminate a falsely negative conclusion, however, precludes the possibility of achieving such statistical assurance. It is unlikely that a definitive experiment can ever settle the question. Indeed, our investiga-
544
CHOJKIER ET AL.
tion was terminated because it was impossible to continue for ethical reasons. The random selection of IAV resulted in a greater delay of several hours in starting therapy than when IVV was selected. Once it had become evident that intravenous vasopressin could control hemorrhage, the intrinsic delay in initiating treatment became intolerable to the house staff. Indeed, it may be unethical to perform a study in which a potentially hazardous delay in starting therapy is inherent in one of the types of treatment under study. Conceivably, a valid clinical trial can be designed in which IVV is used in all patients initially for a short period at low dosage, after which patients who are still bleeding can be randomized to either IVV or IAV. Hemorrhage was controlled in half the patients who received either form of therapy. This relatively low control rate is lower than that reported in some previous studies3.5.6.‘4,16.17,21,23 but is in accord with that observed in other series.4.24.2” Transfusion requirements were similar in the two groups (Table 2). The rate of transfusion was reduced slightly by both IVV (0.26-0.20 transfusions per hour; 77%) and IAV (0.360.21 transfusions per hour; 58%). The IAV group appears to have had more severe bleeding based on transfusion requirements before randomization, but this difference is not statistically significant. The duration of vasopressin therapy was similar in the two groups (18 vs. 16 hr). The dosage of vasopressin administered was significantly greater with IVV than with IAV (540 vs. 221 U). This two- to three-fold difference reflects the higher concentrations of vasopressin infused systemically. These dosage levels were determined arbitrarily to compensate for the greater volume of distribution and the greater amount of degradation of systemically administered vasopressin as shown by plasma levels of vasopressin administered by various routes (CA Athanasoulis, personal communication). There are several differences in the two controlled comparisons of IVV and IAV. Johnson et al. used the same dosage (0.4 U/min) both intraarterially and intravenously. This initial intraarterial dose is larger than that usually given. We began each patient with a much smaller intraarterial dose (0.1 U/min) and increased it as needed. Our initial intravenous dose (0.3 U/min), which was similar to theirs, was increased as needed to much higher dosages (up to 1.5 U/min). The fact that bleeding was controlled as frequently and effectively with the arbitrary intravenous dose as with the customary intraarterial dose, and that the nature, frequency, and severity of systemic side-effects and complications were similar with the two routes of administration, indicates that the IVV doses we employed were the pharmacologic equivalent of conventional intraarterial doses.
GASTROENTEROLOGY
Vol. 77, No. 3
100
6
75
%
50
25
0
-
-
IAV Figure 1. Comparison of intravenous and intraarterial vosopressin in the control of voriceal hemorrhage. The crosshatched areas indicate the percentage of patients in whom bleeding was controlled. The numbers indicate the number of patients in each subgroup. Control of hemorrhage was the same with both forms of therapy.
Another difference is the definition of the criteria for control of hemorrhage. Although Johnson’s results show a higher control rate for both IVV and IAV, these differences are not statistically significant (Table 3). The most discouraging finding in our investigation was the inordinately low survival rate, which is
9
50
25
IAV Figure 2. Comparison of intravenous and intraarterial vasopressin on survival in variceal hemorrhage. The crosshatched areas indicate the percentage of patients who survived. The numbers indicate the number of patients in each subgroup. The difference in survival is not statistically significant.
September
1979
Table 3.
Comparison
INTRAARTERIAL
of Results of Two Controlled Control
Investigation
Johnson
et al”
No. of patients
11
No.
7
AND
INTRAVENOUS
VASOPRESSIN
Trials of IVV and IAV of hemorrhage %
INTRAVENOUS 64
Significance”
No.
Survival __~_ % --__
et al
10
5
Johnson
et al.”
14
10
50 INTRAARTERIAL 71
6
55
3
30
10
71
3
25
16
64
6
27
NS
VASOPRESSIN P < 0.05
NS Chojkier
et al
12
6
Johnson
et al”
25
17
50 TOTAL 68
GROUP P < 0.02
NS Chojkier
et al
” NS = Not statistically
22
11
Significance”
VASOPRESSIN NS
ChoIkier
545
IN GI HEMORRHAGE
50
significant.
lower than that observed by Johnson et al.” Using the same criteria” in patients with bleeding varices, our survival rate is not significantly different for IVV but is significantly lower for IAV and for the total group (Table 3). The explanation for this difference is not evident, because the two groups are comparable in clinical and laboratory manifestations of cirrhosis and in the number of units of blood administered before randomization. Despite the similarity of results by the two routes of administration of vasopressin on control of hemorrhage and on survival, the possibility of type II errors (falsely negative conclusions) must be considered. Because control of hemorrhage was achieved in 50% with intravenous and 50% with intraarterial the number of patients required to vasopressin, achieve statistical significance would be infinite, an unlikely goal for a finite investigation. If the present survival trends were to continue (3 of 10 for IVV; 3 of 12 for IAV), to achieve statistical significance at the 5% level for survival would require approximately 1250 patients. At the present rate of accessions, this would take an additional 83 yr to complete, an extraordinary tenure for ordinary investigations. We conclude, instead, that a type II error is extremely unlikely. One of the most distressing findings of most of the controlled investigations of vasopressin, including the present one, is that there has been no significant improvement in survival, despite effective control of the bleeding.‘-’ These observations suggest that vasopressin by any route may not improve the ultimate outcome and indicate that the use of vasopressin in the treatment of bleeding varices has not been established. It is possible that the systemic side-effects and complications of vasopressin may contribute to the high mortality rate. To avoid such poten-
tial risks, we recommend that a brief (2-4 hr) lowdose (progressively increasing from 0.3 to 0.9 U/min) therapeutic trial of intravenous vasopressin be used early in the treatment of variceal hemorrhage. This recommendation is based on the observations in our studies that of the 4 patients whose hemorrhage was controlled by IVV, 1 was controlled at 0.3 U/min, 1 at 0.6 and 2 at 0.9 U/min, and none was controlled at higher rates when lower infusion rates had failed. None of the 4 had complications. If it is quickly effective, definitive therapy can be elected at the optimal time. If it is not, other methods of stopping the bleeding should be instituted without delay.
References 1. Kehne JH, Hughes FA, Gompertz ML: The use of surgical pituitrin in the control of esophageal varix bleeding: an experimental study and report of two cases. Surgery 39:917,1956 2. Merigan TC, Jr., Plotkin GR, Davidson CS: Effect of intravenously administered posterior pituitary extract on hemorrhage from bleeding esophageal varices. A controlled evaluation. N Engl J Med 268:134,1962 3. Conn HO, Ramsby GR, Storer EH, et al: Intraarterial vasopressin in the treatment of upper gastrointestinal hemorrhage: a prospective, controlled clinical trial. Gastroenterology 68:211,1975 4. Mallory A, Schaefer JW, Cohen JR, et al: Selective intraarterial vasopressin for upper gastrointestinal hemorrhage: a controlled trial. Arch Surg (in press) 5. Johnson WC, Widrich WC, Ansell JE, et al: Control of bleeding varices by vasopressin: a prospective randomized study. Ann Surg 186:369,1977 6. Baum S, Nusbaum M: The control of gastrointestinal hemorrhage by selective mesenteric arterial infusion of vasopressin. Radiology 98:497,1971 7. Baum S, Athanasoulis CA, Waltman AC: Angiographic diagnosis and control of large-bowel bleeding. Dis Colon Rectum 17:447, 1974 8. Athanasoulis CA, Baum S, Waltman AC, et al: Control of acute gastric mucosal hemorrhage. N Engl J Med 290:597,1974
546
CHOJKIER
9. Schwartz
ET AL.
SI, Bales HW, Emerson GL, et al: The use of intravenous pituitrin in treatment of bleeding esophageal varices. Surgery 45:72,1959 10. Shaldon S, Sherlock S: The use of vasopressin (“Pitressin”) in the control of bleeding from oesophageal varices. Lancet 2:222, 1960 11. Conn HO, Dalessio DJ: Multiple infusions of posterior pituitary extract in the treatment of bleeding esophageal varices. Ann Intern Med 57:804,1962 12. Thomford NR, Sirinek KR: Intravenous vasopressin in patients with portal hypertension: Advantages of continuous infusion. J Surg Res 18:113,1975 13. Kaufman SL, Maddrey WC, Harrington DP, et al: Hemodynamic effects of superior mesenteric arterial and intravenous vasopressin infusions in patients with portal hypertension. Invest Radio1 12:210,1977 14. Rigberg LA, Ufberg MH, Brooks CM: Continuous low dose peripheral vein pitressin infusion in the control of variceal bleeding. Am J Gastroenterol68:481, 1977 AR, Kerr JC, Swan KG, et al: Primate mesenteric 15. Freedman blood flow. Effects of vasopressin and its route of delivery. Gastroenterology 74:875,1978 16. Conn HO, Ramsby GR, Storer EH: Selective intraarterial vasopressin the treatment of upper gastrointestinal hemorrhage. Gastroenterology 63:634,1972 17. Marubbio AT, Lombard0 RP, Holt PR: Control of variceal bleeding by superior mesenteric artery pitressin perfusions. Complications and indications. Am J Dig Dis 18539, 1973 18. Marubbio AT: Antidiuretic hormone effect of pitressin during continuous pitressin infusion. Gastroenterology 62:1103,1972
GASTROENTEROLOGY
Vol. 77, No. 3
ef19. Millette B, Huet P-M, Lavoie P. et al: Portal and systemic fects of selective infusion of vasopressin into the superior mesenteric artery in cirrhotic patients. Gastroenterology 69:6, 1975 venous 20. Renert WA, Button KF, Fuld SL, et al: Mesenteric thrombosis and small-bowel infarction following infusion of vasopressin into the superior mesenteric artery. Radiology 102:299, 1972 21. Johnson WC, Widrich WC: Efficacy of selective splanchnic arteriography and vasopressin perfusion in diagnosis and treatment of gastrointestinal hemorrhage. Am J Surg 131:481, 1976 22. Berardi RS: Vascular complications of superior mesenteric artery infusion with pitressin in treatment of bleeding esophageal varices. Am J Surg 127:757.1974 23. Kaufman SL, Harrington DP, Barth KH, et al: Control of variteal bleeding by superior mesenteric artery vasopressin infusion. Am J Roentgen01 128:567, 1977 24. Murray-Lyon IM, Pugh RNH, Nunnerley HB, et al: Treatment of bleeding oesophageal varices by infusion of vasopressin into the superior mesenteric artery. Gut 14:59,1973 LC, Brink RR, Wolfman EF: Survival 25. Getzen fusion of pitressin into the superior mesenteric trol bleeding esophageal varices in cirrhotic Surg 187:337,1978
following inartery to conpatients. Ann
RA, Rheingold OJ, Chiprut RO, et al: Local gan26. Greenwald grene: a complication of peripheral pitressin therapy for bleeding esophageal varices. Gastroenterology 74:744,1978
77:540-546,1979
A Controlled Comparison of Continuous Intraarterial and Intravenous Infusions of Vasopressin in Hemorrhage from Esophageal Varices M. CHOJKIER, S. BAR-MEIR, J. L. KNIAZ,
R. J. GROSZMANN, C. E. ATTERBURY, A. T. BLEI, J. FRANKEL, M. G. GLICKMAN, R. SCHADE,
G. J. TAGGART,
and H. 0.
CONN
Liver Research Laboratory, Veterans Administration Medical Center, West Haven, Connecticut, and The Department of Internal Medicine, Yale University School of Medicine, New Haven, Connecticut
Infusions of intraarterial vasopressin (IAV) into the superior mesenteric artery have been shown to be effective in controlling hemorrhage from esophagogastric varices. Intravenous infusions of vasopressin (NV), which can be initiated rapidly and require less sophisticated equipment and personnel, have also been reported to control variceal hemorrhage. We undertook a controlled clinical trial to compare these two routes of administration. Twenty-two cirrhotic patients with massive hemorrhage from varices were randomized to receive either IVV or IAV. Intraarterial vasopressin was begun at 0.2 U/min and increased progressively as needed to 0.2,0.3, 0.4, and 0.5 U/min. Intravenous vasopressin was begun at 0.3 U/min and increased progressively as needed to 0.6,0.9,1.2,and 1.5 U/min. Hemorrhage was controlled in 5 of 10 episodes (50%) with IVV and in 6 of 1.2episodes (50%) with IAV. Seven of the ten episodes treated with IVV (70%) ended fatally compared with 9 of 12 treated with IAV (75%). Side-effects and complications occurred with similar frequency in the two groups. The two routes of administration are equal in effects, side-effects, and complications. We recommend that NV, which can be administered more easily, be given a brief therapeutic trial early in the management of hemorrhage from varices.
Posterior substance,
pituitary extract, a potent vasoconstrictive was reported in 1956 to be useful in the
Received January 3,1979. Accepted May 11,1979. Address requests for reprints to: H. 0. Conn, M.D., VA Medical Center, West Haven, Connecticut 06516. 0 1979 by the American Gastroenterological Association 0016-5085/79/090540-07$02.00
treatment of bleeding esophageal varices.’ During the succeeding 20 yr, controlled clinical trials have shown that posterior pituitary extract, or its vasoactive principle vasopressin, is effective in controlling hemorrhage from esophageal varices. First, Merigan et al. showed that intravenous, bolus injections were successful.’ Subsequently, two groups found that continuous infusions of vasopressin into the superior mesenteric artery controlled variceal bleeding.“,’ Most recently, Johnson et al. reported that continuous intravenous vasopressin was as effective as intraarterial infusions in the treatment of hemorrhage from varices.’ Several of these report? and others’-* suggested that vasoconstrictive therapy might be effective in nonvariceal bleeding lesions as well. Continuous infusions of vasopressin into the superior mesenteric artery had been introduced to decrease portal pressure while avoiding the systemic effects observed with the bolus intravenous injection of vasopressin.“-” Recent studies have suggested that the hemodynamic responses, complications, and the therapeutic effects of intravenous vasopressin infusions”.‘2-‘5 were similar to those of intraarterial infusions.“-G.‘“~ZG Indeed, Merigan and his colleagues at the Boston City Hospital showed in the first prospective randomized investigation of vasoconstrictive therapy that although the bleeding was controlled, the mortality rate was not significantly decreased.’ Unfortunately, this early experience appeared to set the pattern for subsequent studies.“-’ The simplicity, speed, and apparent success of the systemic intravenous administration of vasopressin, which requires neither experienced angiographers nor expensive, sophisticated equipment, suggest that this route of administration might be more practical
September 1979
INTRAARTERIAL AND INTRAVENOUS VASOPRESSIN IN GIHEMORRHAGE
than selective intraarterial infusion. The only published controlled comparison of intravenous and intraarterial vasopressin, which showed the two routes to be equally effective, was performed in a relatively small group of patients.” The present investigation was undertaken to compare the effect of selective intraarterial infusions of vasopressin (IAV) with systemic intravenous vasopressin (NV) on the control of upper gastrointestinal hemorrhage, on survival, and on the frequency, nature, and severity of the systemic complications of vasopressin therapy.
Methods and Materials Patients with massive upper gastrointestinal (GI) bleeding admitted to the West Haven Veterans Administration Hospital between May 1976 and December 1977 were candidates for inclusion in this investigation. Massive hemorrhage was defined as the loss of 1000 ml or more
blood,
confirmed
by a drop in hematocrit of at least of vital signs (pulse rate of > lOO/ min, blood pressure < 100/60mm Hg, or a decrease in blood pressure of > 20/10mm Hg on the patient’s assuming the sitting position). Patients with known coronary or cerebral artery disease were excluded. After informed consent had been obtained, patients were randomized by sealed envelope technique to receive either IVV or IAV infusions in addition to conventional medical management. The randomization scheme was devised by Mr. E. Lerner, a biostatistician, whose assistance is acknowledged, using a table of random numbers. Decisions about eligibility were made by the postdoctoral fellows of the Liver Research Laboratory and the responsible house officer and confirmed by the attending physician before randomization. Conventional medical management is defined as any form of treatment except vasoconstrictive therapy. Separate randomization schemes were used for patients bleeding from esophagogastric varices and for those bleeding from other, nonvariceal sites. This investigation was approved by the Human Investigations Committee on January 23,1976. The nature of the investigation and the potential dangers of vasopressin infusion were explained to the patients and/or their next of kin, and written authorization for participation in the investigation was obtained. The decisions about the control of hemorrhage in individual patients were made retrospectively on prospectively recorded data by two of the investigators (MC and HOC) independently. The investigators, who had no preconceived opinions and who made these decisions solely on the basis of predetermined criteria, agreed in each instance. Student’s paired and unpaired t-tests and x2 tests were used in statistical analyses. Only seven episodes of GI hemorrhage from nonvariceal sources were randomized, a number too small to evaluate, and, consequently, this report deals primarily with those patients who bled from varices. Twenty-two episodes of hemorrhage from varices in 19 patients satisfied these criteria and were randomized in this investigation. All were men who ranged in age from
6% and by instability
541
25 to 77 (mean 54) yr of age. All were cirrhotic patients except one, who was shown subsequently to have a portal vein thrombosis. Attempts were made to establish the site of hemorrhage by endoscopy and/or arteriography in virtually all. Endoscopy was performed in 20 of the 22 episodes (91%). In one patient, previous endoscopic examinations had shown esophageal varices, and arteriography was done instead. Another patient developed shock before endoscopy could be performed, and randomization was carried out without a definitive diagnosis. Arteriography (64%).In five epiwas performed in 14 of the 22 episodes sodes, the bleeding site had been identified by endoscopy, and arteriography was considered unnecessary. In one instance, the patient developed shock and died before angiography could be undertaken. In 2 patients, the bleeding stopped spontaneously, and no further studies were performed. The site of hemorrhage was considered proved if the bleeding lesion was demonstrated at endoscopy, probable, if varices were identified in the absence of any other potential bleeding lesion, and clinical if not so clearly established. Two patients who had massive variceal hemorrhage were excluded from this investigation. One was excluded because of coronary artery disease, and he subsequently died. The other was excluded because the bleeding had stopped spontaneously; he survived. The site of hemorrhage was proved in 7, probable in 11, and clinical in 4 of the 22 episodes. Ten patients were randomized to receive IVV and 12 to receive IAV. Three patients were randomized twice, each on separate hospital admissions. One of them was selected for IVV on both occasions, one for IAV twice, and the third was treated once by each route. Each patient had a nasogastric tube in place, and serial gastric aspirations were obtained. Hemorrhage was considered controlled by vasopressin if bleeding was stopped for a period of at least 24 consecutive hr, as evidenced by the absence of blood in the gastric aspirations and by stability of vital signs and hematocrit. Vasopressin was administered continuously with a Harvard pump at a concentration of 1 Unit/ml of physiologic saline. Intraarterial infusions were begun at 0.1 U/min for at least 30 min, and if ineffective, the dosage was progressively increased at 30- to 60-min intervals to 0.2, 0.3, 0.4, and 0.5 U/min. Intravenous infusions were begun at 0.3 U/min for at least 30 min and, if ineffective, were progressively increased at approximately 30- to 60-min intervals to 0.6, 0.9, 1.2, and 1.5 U/min. The duration of hemorrhage before the institution of therapy was calculated from the time the bleeding began, usually outside of the hospital, until the time vasopressin was begun. All patients were treated in an Intensive Care Unit with continuous cardiac monitoring. If treatment was unsuccessful, depending on the rate of bleeding and the clinical situation, therapy was declared a failure and the patient could thereafter be treated with the other route of vasopressin administration, which is hereafter referred to as crossover therapy. Esophageal tamponade was used in
only 1 randomized patient. The balloon tube was passed while awaiting arterial catheterization. Recurrent hemorrhage was defined as renewed bleeding
CHOJKIER ET AL.
542
Table 1.
Comparsion
GASTROENTEROLOGY
of Laboratory
Values in WV and IAV Groups
Observation No. of episodes Age (~4 Duration of hemorrhage (hr) Hematocrit at onset of therapy (%) Hemoglobin at onset of therapy (g%) Ascites (a) Encephalopathy (%) Serum albumin (g%) Blood urea nitrogen (mg%) Total serum bilirubin (mg %) Direct serum bilirubin (mg %) Serum alkaline phosphatase (ILJ) Serum aspartic aminotransferase (IU) Total serum protein (g%) Prothrombin
time prolongation
Data expressed
IAV group
Statistical significance”
10
12 52 f 7 15 f 11 29f5 10.2-+1.5 50 17 3.1+ 0.5 31f25 6.2 -t5.6 3.7+ 4.3 158 f 77 63f47 6.7 + 1.1 3.4f 2.6
NS NS
Results Comparison
of Treatment
Groups
The groups selected for IVV and for IAV were similar at the time of randomization in age, diagnosis, clinical findings, and laboratory values (Table 1). They were also similar in the severity of hemorrhage as manifested by the duration of the hemorrhage, the hematocrit and hemoglobin concentrations, and the number of transfusions administered at the time of randomization and at the time therapy started (Tables 1 and 2). The mean interval between randomization and the initiation of therapy was 45 min for IVV and 3 hr for IAV. Comparison Therapeutic
NS
NS NS NS NS NS NS NS NS NS NS NS
Control
of Hemorrhage
Bleeding was controlled with IVV in 5 of 10 episodes (50%) randomized to IVV and in 6 of 12 episodes (50%) selected for IAV (Figure 1). In 3 patients selected for IAV therapy and in 1 randomized for IVV, bleeding stopped before vasopressin was administered. In one instance, a patient randomized to receive IAV died before vasopressin therapy was begun. Inclusion or exclusion of these nonconforming patients does not alter the results appreciably. Thus, bleeding was actually controlled by the administration of IVV in 4 of 9 episodes and by IAV in 3 of 8 episodes. Nonvariceal bleeding was controlled in 1 of 3 patients treated with IVV and in 1 of 4 with IAV. Crossover therapy was employed on nine occasions. In 5 episodes, IVV failed to control the hemorrhage and IAV was tried. In 3 of the 5 (60%), therapy with IAV was successful. In 4 episodes, IAV failed to control hemorrhage and IVV was then used. In none of the 4 was IVV treatment successful. These
of WV and ZAV in the Treatment of Variceal Hemorrhage index
No. of episodes Control of hemorrhage Survival Duration of vasopressin therapy (hr) Amount of vasopressin infused (U) Transfusion prior to vasopressin (U) Transfusions with vasopressin (U) Rate of transfusions prior to vasopressin (U/hr) Rate of transfusions with vasopressin (U/hr) Major complications Minor complications Data expressed
NS
as mean + SD. n NS = Not significant.
within seven days of the successful treatment of bleeding varices. Survival was defined as the discharge of the patient, alive, from the hospital during the admission in which the bleeding occurred.
Table 2.
at Time of Randomization
IVV group 56 f 15 16 f 11 30 +3 10.3-t1.2 50 20 3.2-c0.7 32 +14 3.8-c3.7 3.2f 6.2 138r?:72 190+281 6.3 -+1.3 3.3+ 3.3
(set)
Vol. 77, No. 3
IVV
IAV
5(50%) 3(30%) 18 +ll 54O-t285 4.3 -c1.2 3.7+ 3.8 0.26 0.20 3 4
12 6 (50%) 3(25%) 16 +lO 221*135 5.4f 3.0 3.4 f 4.6 0.36 0.21 3 2
10
as Mean f SD. *P c 0.02 (dosage determined by treatment allocation) 0 NS = Not significant.
Statistical significance” NS NS NS * NS NS NS NS
NS NS
Scptembcr
INTRAARTERIAL
1979
AND INTRAVENOUS VASOPRESSIN IN Gl HEMORRHAGE
data suggest that IAV may be more effective than IVV when the “other” route of administration has failed. Recurrent hemorrhage occurred after successful therapy with IVV in 2 of 5 patients (40%). In 1 of these patients, IVV was again used and was successful. Recurrent bleeding was observed in 3 of the 6 patients (50%) controlled with IAV. In only 1 of the 3 patients was bleeding again stopped by IAV. These differences are not statistically significant.
Survival Seven of the 10 episodes (70%) treated with IVV ended fatally, compared with 9 of the 12 (75%) treated with IAV. This difference is not statistically significant (Figure 2). Two of 3 patients with nonvariceal bleeding treated with IVV died, as did all 4 who received IAV.
Duration
of Therapy
Before admission to the study, the IVV and IAV groups had been bleeding for a mean of 16 and 15 hr, respectively. In the IVV group, the duration of therapy ranged from 5 to 32 hr and averaged 18 hr. In the IAV group, the superior mesenteric artery was infused. Vasopressin was infused for periods of 3 to 27 hr, with a mean of 16 hr. The catheter remained in place, including saline infusions, for an average of 32 hr, and one remained in place for 97 hr. In 1 patient in whom two separate catheterizations were performed, the total duration of catheterization was 121 hr. More vasopressin was infused into the patients by the intravenous route (mean, 540 U) than by the intraarterial route (mean 221 U), a significant difference (P < 0.02). This difference reflects the higher concentration of vasopressin given intravenously.
Transfusion
Requirements
Before admission to the study, the IVV group received an average of 4 U of blood compared with 5 U in the IAV group. Thus, the patients in the two groups had received 0.26 and 0.36 U of blood per hour of hemorrhage, respectively (Table 2). These differences are not statistically significant. After vasopressin was begun, the IVV group received an average of 4 U of blood compared with 3 U with the IAV group (Table 2). Thus, during vasopressin therapy, the IVV group received 0.20 U/hr and the IAV group was transfused with 0.21 U/hr. Neither of these differences is significant.
543
Complications Among the IVV group, three patients had major complications. One had a cardiorespiratory arrest after 10 hr of treatment at a maximal rate of 0.9 U/min. An idioventricular rhythm of 30 per minute and extreme acidosis (pH 6.95) were corrected, but bleeding persisted, and the patient died of hemorrhage 12 hr later. A second patient had phlebitis at the site of the infusion and staphylococcal septicemia. Antibiotic therapy was successfully administered. The third complication in the IVV group may actually have been caused by IAV therapy. After IVV had been administered for 10 hr at a rate up to 1.5 U/min without controlling hemorrhage, IAV was administered for 10 additional hr, at which time the patient developed shock. The patient died 20 hr after IAV had been stopped, and autopsy showed ischemic necrosis of the large bowel. Four additional patients had minor complications such as hypertension, bradycardia, or ventricular extrasystoles. Among the IAV group, 3 patients had major complications and 1 patient developed bacteremia, with a Klebsiella species and Staphylococcus aureus. The staphylococcus was also cultured from the tip of the catheter. One patient developed acute pulmonary edema after 1 hr of successful vasopressin administration at the rate of 0.3 U/min. Vasopressin was stopped and the cardiac status improved, but when bleeding recurred, IAV was again started and pulmonary arterial wedged pressure rose alarmingly, and IAV was again discontinued. One patient bled massively (estimated blood loss of 1000 ml) from the site of a prolonged and difficult catheterization. Two other patients experienced clotted catheters, which required replacement.
Discussion This investigation supports the previous observations of Johnson et al. that continuous intravenous infusions of vasopressin are as effective in the treatment of bleeding varices as infusions of vasopressin into the superior mesenteric artery.s It also confirms numerous reports that the same systemic side-effects and complications occur with IAV as with intravenous administration.“~“,‘~-‘~,~l-~~ Unfortunately, the two controlled clinical trials together include only 47 patients, far too few to exclude the possibility that one of these routes of administration is truly superior to the other. The large number of additional patients that would be needed to eliminate a falsely negative conclusion, however, precludes the possibility of achieving such statistical assurance. It is unlikely that a definitive experiment can ever settle the question. Indeed, our investiga-
544
CHOJKIER ET AL.
tion was terminated because it was impossible to continue for ethical reasons. The random selection of IAV resulted in a greater delay of several hours in starting therapy than when IVV was selected. Once it had become evident that intravenous vasopressin could control hemorrhage, the intrinsic delay in initiating treatment became intolerable to the house staff. Indeed, it may be unethical to perform a study in which a potentially hazardous delay in starting therapy is inherent in one of the types of treatment under study. Conceivably, a valid clinical trial can be designed in which IVV is used in all patients initially for a short period at low dosage, after which patients who are still bleeding can be randomized to either IVV or IAV. Hemorrhage was controlled in half the patients who received either form of therapy. This relatively low control rate is lower than that reported in some previous studies3.5.6.‘4,16.17,21,23 but is in accord with that observed in other series.4.24.2” Transfusion requirements were similar in the two groups (Table 2). The rate of transfusion was reduced slightly by both IVV (0.26-0.20 transfusions per hour; 77%) and IAV (0.360.21 transfusions per hour; 58%). The IAV group appears to have had more severe bleeding based on transfusion requirements before randomization, but this difference is not statistically significant. The duration of vasopressin therapy was similar in the two groups (18 vs. 16 hr). The dosage of vasopressin administered was significantly greater with IVV than with IAV (540 vs. 221 U). This two- to three-fold difference reflects the higher concentrations of vasopressin infused systemically. These dosage levels were determined arbitrarily to compensate for the greater volume of distribution and the greater amount of degradation of systemically administered vasopressin as shown by plasma levels of vasopressin administered by various routes (CA Athanasoulis, personal communication). There are several differences in the two controlled comparisons of IVV and IAV. Johnson et al. used the same dosage (0.4 U/min) both intraarterially and intravenously. This initial intraarterial dose is larger than that usually given. We began each patient with a much smaller intraarterial dose (0.1 U/min) and increased it as needed. Our initial intravenous dose (0.3 U/min), which was similar to theirs, was increased as needed to much higher dosages (up to 1.5 U/min). The fact that bleeding was controlled as frequently and effectively with the arbitrary intravenous dose as with the customary intraarterial dose, and that the nature, frequency, and severity of systemic side-effects and complications were similar with the two routes of administration, indicates that the IVV doses we employed were the pharmacologic equivalent of conventional intraarterial doses.
GASTROENTEROLOGY
Vol. 77, No. 3
100
6
75
%
50
25
0
-
-
IAV Figure 1. Comparison of intravenous and intraarterial vosopressin in the control of voriceal hemorrhage. The crosshatched areas indicate the percentage of patients in whom bleeding was controlled. The numbers indicate the number of patients in each subgroup. Control of hemorrhage was the same with both forms of therapy.
Another difference is the definition of the criteria for control of hemorrhage. Although Johnson’s results show a higher control rate for both IVV and IAV, these differences are not statistically significant (Table 3). The most discouraging finding in our investigation was the inordinately low survival rate, which is
9
50
25
IAV Figure 2. Comparison of intravenous and intraarterial vasopressin on survival in variceal hemorrhage. The crosshatched areas indicate the percentage of patients who survived. The numbers indicate the number of patients in each subgroup. The difference in survival is not statistically significant.
September
1979
Table 3.
Comparison
INTRAARTERIAL
of Results of Two Controlled Control
Investigation
Johnson
et al”
No. of patients
11
No.
7
AND
INTRAVENOUS
VASOPRESSIN
Trials of IVV and IAV of hemorrhage %
INTRAVENOUS 64
Significance”
No.
Survival __~_ % --__
et al
10
5
Johnson
et al.”
14
10
50 INTRAARTERIAL 71
6
55
3
30
10
71
3
25
16
64
6
27
NS
VASOPRESSIN P < 0.05
NS Chojkier
et al
12
6
Johnson
et al”
25
17
50 TOTAL 68
GROUP P < 0.02
NS Chojkier
et al
” NS = Not statistically
22
11
Significance”
VASOPRESSIN NS
ChoIkier
545
IN GI HEMORRHAGE
50
significant.
lower than that observed by Johnson et al.” Using the same criteria” in patients with bleeding varices, our survival rate is not significantly different for IVV but is significantly lower for IAV and for the total group (Table 3). The explanation for this difference is not evident, because the two groups are comparable in clinical and laboratory manifestations of cirrhosis and in the number of units of blood administered before randomization. Despite the similarity of results by the two routes of administration of vasopressin on control of hemorrhage and on survival, the possibility of type II errors (falsely negative conclusions) must be considered. Because control of hemorrhage was achieved in 50% with intravenous and 50% with intraarterial the number of patients required to vasopressin, achieve statistical significance would be infinite, an unlikely goal for a finite investigation. If the present survival trends were to continue (3 of 10 for IVV; 3 of 12 for IAV), to achieve statistical significance at the 5% level for survival would require approximately 1250 patients. At the present rate of accessions, this would take an additional 83 yr to complete, an extraordinary tenure for ordinary investigations. We conclude, instead, that a type II error is extremely unlikely. One of the most distressing findings of most of the controlled investigations of vasopressin, including the present one, is that there has been no significant improvement in survival, despite effective control of the bleeding.‘-’ These observations suggest that vasopressin by any route may not improve the ultimate outcome and indicate that the use of vasopressin in the treatment of bleeding varices has not been established. It is possible that the systemic side-effects and complications of vasopressin may contribute to the high mortality rate. To avoid such poten-
tial risks, we recommend that a brief (2-4 hr) lowdose (progressively increasing from 0.3 to 0.9 U/min) therapeutic trial of intravenous vasopressin be used early in the treatment of variceal hemorrhage. This recommendation is based on the observations in our studies that of the 4 patients whose hemorrhage was controlled by IVV, 1 was controlled at 0.3 U/min, 1 at 0.6 and 2 at 0.9 U/min, and none was controlled at higher rates when lower infusion rates had failed. None of the 4 had complications. If it is quickly effective, definitive therapy can be elected at the optimal time. If it is not, other methods of stopping the bleeding should be instituted without delay.
References 1. Kehne JH, Hughes FA, Gompertz ML: The use of surgical pituitrin in the control of esophageal varix bleeding: an experimental study and report of two cases. Surgery 39:917,1956 2. Merigan TC, Jr., Plotkin GR, Davidson CS: Effect of intravenously administered posterior pituitary extract on hemorrhage from bleeding esophageal varices. A controlled evaluation. N Engl J Med 268:134,1962 3. Conn HO, Ramsby GR, Storer EH, et al: Intraarterial vasopressin in the treatment of upper gastrointestinal hemorrhage: a prospective, controlled clinical trial. Gastroenterology 68:211,1975 4. Mallory A, Schaefer JW, Cohen JR, et al: Selective intraarterial vasopressin for upper gastrointestinal hemorrhage: a controlled trial. Arch Surg (in press) 5. Johnson WC, Widrich WC, Ansell JE, et al: Control of bleeding varices by vasopressin: a prospective randomized study. Ann Surg 186:369,1977 6. Baum S, Nusbaum M: The control of gastrointestinal hemorrhage by selective mesenteric arterial infusion of vasopressin. Radiology 98:497,1971 7. Baum S, Athanasoulis CA, Waltman AC: Angiographic diagnosis and control of large-bowel bleeding. Dis Colon Rectum 17:447, 1974 8. Athanasoulis CA, Baum S, Waltman AC, et al: Control of acute gastric mucosal hemorrhage. N Engl J Med 290:597,1974
546
CHOJKIER
9. Schwartz
ET AL.
SI, Bales HW, Emerson GL, et al: The use of intravenous pituitrin in treatment of bleeding esophageal varices. Surgery 45:72,1959 10. Shaldon S, Sherlock S: The use of vasopressin (“Pitressin”) in the control of bleeding from oesophageal varices. Lancet 2:222, 1960 11. Conn HO, Dalessio DJ: Multiple infusions of posterior pituitary extract in the treatment of bleeding esophageal varices. Ann Intern Med 57:804,1962 12. Thomford NR, Sirinek KR: Intravenous vasopressin in patients with portal hypertension: Advantages of continuous infusion. J Surg Res 18:113,1975 13. Kaufman SL, Maddrey WC, Harrington DP, et al: Hemodynamic effects of superior mesenteric arterial and intravenous vasopressin infusions in patients with portal hypertension. Invest Radio1 12:210,1977 14. Rigberg LA, Ufberg MH, Brooks CM: Continuous low dose peripheral vein pitressin infusion in the control of variceal bleeding. Am J Gastroenterol68:481, 1977 AR, Kerr JC, Swan KG, et al: Primate mesenteric 15. Freedman blood flow. Effects of vasopressin and its route of delivery. Gastroenterology 74:875,1978 16. Conn HO, Ramsby GR, Storer EH: Selective intraarterial vasopressin the treatment of upper gastrointestinal hemorrhage. Gastroenterology 63:634,1972 17. Marubbio AT, Lombard0 RP, Holt PR: Control of variceal bleeding by superior mesenteric artery pitressin perfusions. Complications and indications. Am J Dig Dis 18539, 1973 18. Marubbio AT: Antidiuretic hormone effect of pitressin during continuous pitressin infusion. Gastroenterology 62:1103,1972
GASTROENTEROLOGY
Vol. 77, No. 3
ef19. Millette B, Huet P-M, Lavoie P. et al: Portal and systemic fects of selective infusion of vasopressin into the superior mesenteric artery in cirrhotic patients. Gastroenterology 69:6, 1975 venous 20. Renert WA, Button KF, Fuld SL, et al: Mesenteric thrombosis and small-bowel infarction following infusion of vasopressin into the superior mesenteric artery. Radiology 102:299, 1972 21. Johnson WC, Widrich WC: Efficacy of selective splanchnic arteriography and vasopressin perfusion in diagnosis and treatment of gastrointestinal hemorrhage. Am J Surg 131:481, 1976 22. Berardi RS: Vascular complications of superior mesenteric artery infusion with pitressin in treatment of bleeding esophageal varices. Am J Surg 127:757.1974 23. Kaufman SL, Harrington DP, Barth KH, et al: Control of variteal bleeding by superior mesenteric artery vasopressin infusion. Am J Roentgen01 128:567, 1977 24. Murray-Lyon IM, Pugh RNH, Nunnerley HB, et al: Treatment of bleeding oesophageal varices by infusion of vasopressin into the superior mesenteric artery. Gut 14:59,1973 LC, Brink RR, Wolfman EF: Survival 25. Getzen fusion of pitressin into the superior mesenteric trol bleeding esophageal varices in cirrhotic Surg 187:337,1978
following inartery to conpatients. Ann
RA, Rheingold OJ, Chiprut RO, et al: Local gan26. Greenwald grene: a complication of peripheral pitressin therapy for bleeding esophageal varices. Gastroenterology 74:744,1978