Diagnostic Radiology
New Techniques for Interruption of Gastroesophageal Venous Blood Flow 1 Raul Pereiras, M.D., Manuel Viamonte, Jr., M.D., Edward Russell, M.D., James LePage, M.D., Patrick White, M.D., and Duane Hutson, M.D. Enlarged gastroesophageal veins were successfully obliterated in 41 patients using embolization with modified autogenous clots and/or Gelfoam, balloon occlusion, iatrogenic perivenous hematoma, sclerosing agents (Sotradecol and Keflin), or a combination of these methods. Thirteen patients were actively bleeding when studied, and the site of bleeding was detected in 4. Surgical exploration of 16 patients and autopsy study of 5 showed persistent obliteration ranging between three weeks and seven months. No major complications requiring reparative surgery were encountered. Gelfoam soaked with Sotradecol is the preferred agent because it provides persistent obliteration of the emboli zed veins. Patients who are acutely bleeding or have done so previously are candidates for selective obliteration of the gastroesophageal veins. Embolism, therapeutic (Portal vein, therapeutic embolization, 9[57].129) • Esophagus, varices, 7 [ 1] .750 • Gastrointestinal tract, hemorrhage • Stomach, varices. 7[2].750
INDEX TERMS:
Radiology 124:313-323, August 1977
I UNDERQUIST and Vang recently reported on their use . . of selective obliteration of the coronary vein in 4 patients with portal hypertension and bleeding gastroesophageal varices (6). We have employed superselective catheterization of tributaries of the portal vein by the transhepatic approach and have applied this procedure to (a) the study and treatment of patients with portal hypertension, particularly those with actively bleeding gastroesophageal varices; (b) the differential diagnosis of posthepatitic jaundice; (c) the diagnosis of pancreatic disease; and (d) hemodynamic and physiological studies in patients with portal hypertension (13).
9 ml of blood) were used in 4 patients and a mixture of Gelfoam and Sotradecol in 38 (Figs. 1-11). Balloon occlusion of the coronary vein was employed in one case (Fig. 12). In 4 patients a perivenous hematoma obliterated the veins (Figs. 2 and 13). Intimal damage was induced prior to the use of thrombin. Keflin 2 was used in 2 patients and hypertonic glucose in 3. We were unable to catheterize the portal vein in 3 patients and the gastroesophageal veins in 6.
Technique (a) Gelfoam Administration: A surgical gelatin sponge {Gelfoam)3 is cut into small pieces and soaked in 3 % Sotradecol." The resulting gel-like suspension is injected slowly by hand using a 1-ml tuberculin syringe. Following 0.5 ml of the mixture, a small amount of contrast medium is administered under fluoroscopic control. From 2 to 6 ml of the mixture may be needed to obliterate the gastroesophageal veins successfully. It is unnecessary and potentially hazardous to obstruct the large veins close to the spleno-portal system, as embolic material may enter the system and obstruct intrahepatic branches of the portal vein and/or migrate into intestinal or pancreatic branches. (b) Catheterization: During the venous phase of indirect portography, the skin of the abdominal wall is marked at the points of anterior and lateral projection of the right branch of the portal vein (usually at the tip of the transverse process of T11 or T 12) and the patency of the vessel is established. We prefer to employ the highest possible point in the midaxillary line as the point of entry, paying careful
MATERIAL AND METHODS
Fifty-one studies were performed in 41 patients with hepatic cirrhosis and portal hypertension. Thirteen were bleeding actively at the time of the procedure. Nine patients were studied twice and one was studied three times. The coronary vein, short gastric veins, or both were superselectively catheterized in all cases. Before beginning the patient studies, horizontal, sagittal, and coronal sections from frozen cadavers were used to demonstrate the anatomical position of the bifurcation of the portal vein in the hilar portion of the liver. It is usually equidistant from the front and back of the body (though this varies with the physique of the patient and pathological conditions such as ascites and advanced liver disease) and generally projects ventrally and to the right of the spine at a distance equivalent to the anteroposterior diameter of L1. Modified autogenous blood clots (1 ml of Amicar plus
1 From the Departments of Radiology (R.P., P.W.) and Surgery (D.H.), Veterans Administration Hospital, the Department of Radiology, Jackson Memorial Hospital (M.V., E.R., J.L.P.), and the Department of Surgery, University of Miami School of Medicine (D.H.), Miami, Fla. Revised version accepted for publication in November 1976. 2 Eli Lilly & Co., Indianapolis, Ind. 3 Upjohn Co., Kalamazoo, Mich. 4 Elkins-Sinn, Inc., Cherry Hill, N.J. sjh
313
314
RAUL PEREIRAS AND OTHERS
August 1977
Fig. 1. CASE J. A and B. Embolization of the coronary vein with Gelfoam and perivenous infiltration of the short gastric veins. Superselective catheterization of the coronary vein reveals markedly enlarged gastroesophageal veins with intense opacification of the wall of the esophagus (arrows). Active bleeding was present, with oozing of contrast material. 0 tip of catheter.
=
attention to the farthest lateral inferior projection of the right costodiaphragmatic sulcus. After infiltrating the skin, subcutaneous plane, and intercostal space with a local anesthetic, a 21 gauge needle 5.1 cm (2 in.) long is used to penetrate the peritoneum and liver capsule while the patient stops breathing. A stab incision is made in the skin using a #11 blade, and the skin and subcutaneous tissues are separated with a straight forceps. A 19 gauge trocar'' 30-60 em long (average, 40 cm) with a plastic sleeve (o.d. 1.25 mm) is advanced close to the upper edge of the lower rib and directed parallel to the plane of the x-ray table. During apnea, the tip of the trocar is quickly directed toward the liver hilus (usually at the level of T11 or T12) under fluoroscopic control. The cannula and perforating needle are quickly removed and the patient is asked to breathe gently. Aspiration of blood followed by injection of contrast material will facilitate localization of a large 5 Becton-Dickinson Co., Rutherford N.J.; Johannah Medical Services, Inc., Minneapolis, Minn,
branch of the portal vein, after which a small J-shaped guide wire (o.d. 0.25 mm) is advanced through the catheter; it may take several passes before the portal vein is entered. Rotation of the guide wire will place it in the splenic, inferior mesenteric, or superior mesenteric vein as desired. If the coronary and/or short gastric veins are to be selectively catheterized, the end of the wire should be preshaped to facilitate entering these vessels (5). We have primarily used stainless steel Kifa guide wires. When the tip of the guide wire is in position, the plastic sleeve or other catheters are advanced over it. We have used catheters made of either polyethylene or Teflon (o.d. 1.35-1.58 mm). If catheters have to be exchanged, the patient is asked to breathe in and out and the portion of the guide wire between the liver and the chest wall is observed fluoroscopically to make sure that it is perpendicular to the chest wall, which is necessary to facilitate penetration with the catheter; technical difficulties may be encountered if the guide wire is obliquely directed toward the liver. Control of respiration is critical in this maneuver. Difficulties can
Vol. 124
NEW TECHNIQUES FOR INTERRUPTION OF GASTROESOPHAGEAL BLOOD FLOW
315
Diagnostic Radiology
Fig. 2. CASE I. A. After selective catheterization of the coronary vein (CV) with Gelfoam, Sotradecol, and thrombin, only small venules remain opacified in the esophagogastric region. B. Selective splenic venogram shows filling of the gastroesophageal veins (arrows) via the short gastric veins. Note that the coronary vein is completely occluded and not opacified from this injection. PV = portal vein; SV = splenic vein; 0 = tip of catheter. C. Scout radiograph shows an area of infiltration surrounding the tip of the guide wire. Although the gastric veins were selectively catheterized, the catheter could not be advanced into the smaller venules. A test injection showed extravasation of contrast material, and approximately 15 ml more was then injected to induce spasm. D. Postembolization venogram shows interruption of flow in the gastroesophageal region with nonfilling of the coronary vein and short gastric veins. PV = portal vein; SV = splenic vein.
also occur in patients with a small, scarred right lobe, such as those with advanced cirrhosis. Still under fluoroscopic control, the catheter is advanced to the level of the splenic hilus to obtain a control portogram after removing the gUide wire. This is followed by selective catheterization of all veins connecting the spleno-portal system with the fundus of the stomach and the distal esophagus. If the liver is hardened and the catheter cannot be advanced over the guide wire, as in patients with marked hepatic fibrosis or small portal vein branches, we suggest that coaxial catheterization be employed. Following completion of the study, we attempt to obliterate the juxtacapsular portion of the needle track to avoid
leakage of bile or blood into the peritoneal cavity. For this purpose, the catheter is withdrawn to a point about 2 cm from the surface of the liver and contrast material is injected slowly to ensure nonfilling of the hepatic veins, which if opacified would preclude embolization of the needle track. If the hepatic veins are visualized, the catheter should be withdrawn a few millimeters farther and the test injection repeated; if not, about 0.2 ml of Gelfoam is extruded with a guide wire. If hepatic venous opacification persists, Gelfoam might embolize into the lung; therefore we recommend that thrombin be used instead. We have studied patients with cholestatic jaundice and cirrhosis (most of whom also had ascites) without encountering major bleeding complications.
316
RAUL PEREIRAS AND OTHERS
Fig. 3. CASE 2. Embolization of the coronary vein with Gelfoam and temporary balloon occlusion of the azygos vein. A. Direct portogram shows hepatofugalflow with filling of the coronary (CV), inferior mesenteric (lMV), and umbilical veins (VV). PV = portal vein; SV = splenic vein. B. Selective catheterization of the coronary vein shows free communication with the azygos system (A). 0 = tip of catheter; SV = splenic vein. C. Temporary balloon occlusion (B) of the azygos vein. D. Following embolization of the coronary vein with Gelfoam, angiogram shows the portal (PV) and splenic veins (SV) with no filling of the gastroesophageal veins. Fig. 4. CASE 2. Late phase of the postocclusion portogram demonstrates hepatofugal flow with filling of the inferior mesenteric (lMV) and superior mesenteric veins (SMV). A few short gastric veins are filled (arrows); the patency of these veins may be the cause of subsequentbleeding. PV = portal vein; 0 = tip of catheter.
August 1977
Vol. 124
NEW TECHNIQUES FOR INTERRUPTION OF GASTROESOPHAGEAL BLOOD FLOW
317
Diagnostic Radiology
Fig. 5. CASE 3. Embolization of the short gastric and coronary veins with Gelfoam. A. Findings on the preliminary direct portogram are deceptive, showing the splenic (SV) and portal veins (PV) without demonstrating the gastroesophageal veins. Band C. Selective catheterization of the short gastric veins (SG). Note the partial drainage toward the left renal vein (arrow). D. Postembolization angiogram shows obstruction of the gastroesophageal veins (arrows). Note the short gastric (SG), splenic (SV), and tip of catheter. inferior mesenteric veins (/MV), not seen on the preembolization angiogram. 0 E. Direct portogram now reveals partial filling of the coronary vein (solid arrow) and some flow through one of the short gastric veins (broken arrow) toward the fundus of the stomach. PV = portal vein; SV = splenic vein.
=
RESULTS
Thirteen patients were studied while actively bleeding, but angiographic proof of the bleeding site was demonstrated in only 4. Bleeding was controlled by embolization in all 13 but recurred in 7: the cause of bleeding was incomplete vascular obliteration in 5, Mallory-Weiss syndrome in 1, and an esophageal ulcer in 1. Hemorrhage occurred one week after the procedure in 1 patient, three weeks afterward in 2, four months later in 1, and six months later in 1. Twenty-eight patients were studied while stable. In 9 we merely obliterated the coronary vein, as the patient was being considered for a distal spleno-renal shunt.
Of the 28 patients who are still alive (not the same group as the 28 just mentioned), 16 had surgery following the examination: 2 of them were lost to follow-up and the other 14 have not exhibited further bleeding. Of the 12 patients who did not have surgery, bleeding recurred in one of the 4 followed up for less than one month, none of the 3 studied for one to three months, one of the 2 followed up for three to six months, and none of the 3 studied for six to twelve months. In 4 patients, superselective catheterization of the coronary vein and short gastric veins was unsuccessful and inadvertent extravasation of contrast material around these vessels was noted. Surprisingly, portography revealed complete interruption of flow (Figs. 1 and 13),
318
RAUL PEREIRAS AND OTHERS
August 1977
Fig. 6. CASE 3. Despite poor visualization on the direct portogram, the coronary vein was selectively catheterized. A. Angiogram shows the coronary vein (CV) with an area of narrowing (solid arrows) which corresponds to spasm at the tip of the catheter. Note the filling of the gastroesophageal veins and the large varix (V) in the region of the distal esophagus (broken arrows). B. Postembolization angiogram demonstrates the coronary vein (Cv) and obliteration of the veins previously visualized in the distal esophagus. C. Final postembolization portogram reveals the splenic (SV) and portal veins (PV), but the gastroesophageal veins are not visible.
probably due to venous spasm and compression by a hematoma. Repeat study showed renewed permeability of these vessels, which were subsequently obliterated with a mixture of Gelfoam and Sotradecol. One of our patients had hepatic cirrhosis, portal hypertension, and partial thrombosis of the splenic and portal veins, with very slow blood flow in the gastroesophageal veins. Balloon occlusion of the coronary vein for 20 minutes following intimal trauma and injection of thrombin effectively interrupted flow (Fig. 12). In our patients, the mean pressure in the spleno-portal veins was 1 mm Hg higher than the wedged hepatic vein pressure. Following venous obliteration, the portal vein pressure is usually 2 mm Hg above preembolization levels. When other large porto-systemic collaterals were demonstrated, very little if any rise in portal vein pressure was noted after embolization. No significant complications were observed in our series. Almost all patients had a small subcapsular hematoma of the liver which resolved spontaneously. This was detected on the direct portogram obtained at the completion of embolization: the costal surface of the right lobe of the liver appeared concave and was occasionally separated from the abdominal wall. One patient had active intraperitoneal bleeding which required 1.4 I (3 pints) of blood but no surgical intervention: he demonstrated abnormal coagulation and was on a respirator. Another patient became anuric for four days following a procedure in which 400 ml of contrast medium was used. Renal function slowly returned to precatheterization levels about two weeks later. In one patient in whom only the coronary vein was embolized, massive bleeding occurred 24 hours later, causing death from bleeding esophageal varices. One patient had postcatheterization bleeding caused by inadvertent puncture of an intercostal artery, which was treated by suturing the vessel. Seven of the patients studied while actively bleeding and
2 of those studied while stable died. The cause of death was hepatoencephalopathy in 2, sepsis in 1, recurrent bleeding in 4, and unknown causes in 2. Four patients were lost to follow-up. DISCUSSION
The radiological treatment of gastrointestinal bleeding has included arterial approaches, such as selective, prolonged infusion of vasopressin (9) or a combination of epinephrine and propanolol (8), and injection of autogenous clots, Gelfoam, etc. into the arteries supplying the bleeding area (12), with varying degrees of success. In patients with portal hypertension, intra-arterial infusion of vasopressin has been used most often (1). The approach described here provides immediate and good results. Indirect portography beforehand is desirable but not necessary, and we have not used it in emergency situations. In order to produce effective, permanent obliteration of the gastroesophageal veins, a small catheter should first be advanced as close as possible to the small branches of the coronary and/or short gastric veins. Induction of intimal damage using a guide wire and injection of thrombin or a sclerosing agent as proposed by Lunderquist and Vang (6) may be an adequate method of obliterating veins. However, we believe that in high-flow situations, or when dealing with several veins giving off numerous branches, injection of embolic material into the main collecting trunk (such as the coronary vein) will be more effective. Assessment of the effectiveness of direct venous embolization requires more experience; however, we have documented persistent obliteration of the coronary and short gastric veins with Gelfoam and Sotradecol three weeks to twelve months after the examination and believe that Gelfoam embolization may give the best results. Esophagography is not a good means of differentiating patent from obliterated veins, as varices detected on the postembolization esophagogram may be totally obliterated.
Diagnostic Radiology
Fig. 7. CASE 4. Selective catheterization of three short gastric veins. A. Direct portogram demonstrates hepatofugal flow. PV = portal vein; SMV = superior mesenteric vein; SV splenic vein; C = collateral vein. X and Y represent two short gastric veins supplying the gastroesophageal region. The broken arrow indicates short gastric veins near the splenic hilus which were responsible for the area of extravasation shown in Figure 7, B. B. Late phase of the direct portogram. Note the esophageal varices ( V) and extravasationof contrast material along the greater curvature of the stomach (arrows). Fig. 8. CASE 4. A and B. Selective injection of a short gastric vein (Y) shows opacification of the gastroesophageal veins. Note the extravasation of contrast material into the stomach (broken arrows). C and D. Postembolization study. Arrows indicate the level of occlusion. SV == splenic vein.
=
319
320
RAUL PEREIRAS AND OTHERS
August 1977
Fig. 9. CASE 4. Selective injection of another short gastric vein (X). A. Preembolizationangiogram reveals prominent veins in the esophagogastric region. IMV inferior mesenteric vein; SV splenic vein. There is minimal extravasation of contrast material (broken.arrow). B. Postembolization angiogram shows nonfilling of the esophagogastric veins (solid arrows). Note the extravasation of contrast material (broken arrows) above and below the projection of the splenic vein (SV), which can be seen because of reflux into the splenic vein and subsequent passage into a third short gastric vein (Z). In the preliminary angiogram the area of extravasation was not marked. IMV = inferior mesenteric vein. C. Direct portogram obtained following selective embolization of two short gastric veins (X and Y)(solid arrows). Note the hepatofugal flow and the third short gastric vein (Z) with extravasation of contrast material (broken arrow) just below the projection of the splenic vein (SV). PV portal vein; SMV superior mesenteric vein; IMV inferior mesenteric vein. D. Late phase of the direct portogram shows marked extravasation of contrast material into the stomach (arrows). Note the opacification of one short gastric vein (Z). PV = portal vein. Fig. 10. CASE 4. ' A and B. Selective catheterization of the third short gastric vein (Z). Note the marked extravasation of contrast material into the stomach (arrows). C and D. Postembolization angiogram shows obliteration of the short gastric veins (solid arrows). Extravasation persists, but to a lesser degree (broken arrow).
=
=
=
If the gastroesophageal veins are not completely obliterated, variceal bleeding may recur. When patients are selected for a distal spleno-renal shunt, the coronary vein is ligatedat the time of surgery, while the short gastric
=
=
veins are left undisturbed to provide another outlet for decompression of the gastroesophageal region. Blood flows from the fundus of the stomach through the short gastric veins into the splenic veins and from them into the
Vol. 124
NEW TECHNIQUES FOR INTERRUPTION OF GASTROESOPHAGEAL BLOOD FLOW
Fig. 11. CASE 4. Final portogram. A. Early phase. Note the complete obstruction of the short gastric veins which were embolized (X, Y, Z) (straight arrows). There is hepatofugal flow with filling of the superior mesenteric (SMV) and inferior mesenteric veins (/MV) as well as a large collateral (C) connected to the splenic vein (SV). Filling of the coronary vein (CV) is seen for the first time (curved arrow), with minimal opacification of the esophagogastric veins. PV portal vein. B. Late phase. There is no extravasation of contrast material into the stomach. Note the opacification of the gonadal vein (GV) and inferior vena cava (IVC), suggesting spontaneous diversion of porto-systemic flow. The lack of evidence of active bleeding into the stomach and the considerable amount of contrast material used in performing the procedure explain why the coronary vein was not embolized and the study was discontinued. The patient died of hepatorenal syndrome eight days later. Numerous superficial gastric ulcers were present, but there was no evidence of recurrent bleeding at the esophagogastric junction.
=
321
Diagnostic Radiology
322
RAUL PEREIRAS AND OTHERS
August 1977
Fig. 12. CASE 5. A and B. Balloon occlusion of the coronary vein. Direct portogram revealed occlusion of the superior mesenteric and splenic veins (solid arrows) with hepatofugal flow (broken arrow) via the coronary vein (CV). The slow flow in the coronary vein motivated the use of balloon occlusion. PV = portal vein; 0 = tip of catheter; LPV = left portal vein. C. After inducing intimal damage with a guide wire and injecting thrombin through the catheter, a balloon catheter (B) was inflated in the coronary vein. D. After 15 minutes of balloon occlusion, total interruption of flow in the coronary vein was observed on a control angiogram. PV = portal vein; CV = coronary vein.
left renal vein. In such cases we have obliterated only the coronary vein. We recommend that surgery be performed shortly after transhepatic obliteration because of the possibility of recurrent bleeding from the patent short gastric veins. Patientstreated only with embolization shouldbe studied again four to eight weeks later. If adequate spontaneous porto-systemic communications such as spleno-renal or mesocaval vessels are lacking, other short gastric veins may enlarge and promote further variceal bleeding. In such cases, embolization should be repeated until no gastroesophageal veins are demonstrated angiographically. A potential hazard of this technique is passage of clots or Gelfoam into the azygos system and from there into the pulmonary circulation via the right side of the heart. Temporary balloon occlusion of the azygos vein may prevent this. However, pulmonary perfusion studies made 24 hours after the procedure in 12 patients studied without temporary balloon occlusion of the azygos vein and in 5 autopsies failed to demonstrate pulmonary embolism.
Contraindications include a bleeding tendency and the presence of a hypervascular mass in the path of the catheter. Fever, hypersensitivity to contrast agents, ascites, anemia, and the need for a respirator are considered risk factors but not contraindications. Vital signs must be monitored closely for at least eight hours after the procedure. In cadaver and cast-corrosion studies of patients with cirrhosis of the liver and portal hypertension, we have observed intense vascular engorgement of the spleen contrasted with a reduced hepatic venous bed; these findings coupled with the friability of this organ contraindicate transsplenic catheterization of the portal vein. We have not used the cumbersome transjugular (10) and translumbar (3) approaches. Catheterization of the umbilical portal vein (4, 7) requires extraperitoneal dissection, and we have used it in only a few patients. Transhepatic portography is not a new technique (2, 14). However, selective catheterization of the portal vein and its tributaries for diagnosis and treatment is a new devel-
Vol. 124
NEW TECHNIQUES FOR INTERRUPTION OF GASTROESOPHAGEAL BLOOD FLOW
323
Diagnostic Radiology
Fig. 13. CASE 6. Perivenous infiltration of the coronary vein in a patient with thrombosis of an earlier mesocaval shunt. A. Direct portogram showed hepatofugal flow with filling of the superior mesenteric (SMV), inferior mesenteric (lMV), and coronary veins (CV). Contrast material injected by hand in an attempt to advance the coronary vein catheter into smaller veins appeared to extravasate; nevertheless, approximately 15 ml was injected. PV portal vein; SV splenic vein; arrows indicate the direction of venous flow. B. A second injection was carried out with the tip of the catheter in the splenic vein (0). Note the extravasated contrast material in the gastroesophageal region (broken arrows); flow is spontaneously diverted into the left renal vein (LRV) and from there to the inferior vena cava (lVC) through large retroperitoneal veins which empty into the left spermatic and left adrenal veins (solid arrows). C. A third injection was carried out with the tip of the catheter (0) at the junction of the splenic (SV) and portal veins (PV). Broken arrows indicate interruption of flow in the coronary vein secondary to perivascular extravasation of contrast material; solid arrows indicate the direction inferior mesenteric vein. of venous flow. SMV = superior mesenteric vein; IMV
=
=
=
opment (6, 13). When the platelet count is below 40,000 or the prothrombin level is 7 over control, the umbilical approach appears to be safest. The transjugular technique has been proposed in such situations and also when ascites is present (10). In our patients with ascites (80 % ), we used the transhepatic approach only with hesitation. None required surgical intervention following the procedure or medical treatment for bleeding.
SUMMARY Bleeding from ruptured esophageal varices is the most common cause of death in patients with cirrhosis and portal hypertension. We believe that transhepatic catheterization and obliteration of the gastroesophageal veins with Gelfoam and Sotradecol is a safe and effective method of halting or preventing bleeding in such cases. ADDENDUM: Since this paper was submitted, we have studied more than 80 patients, with follow-up for up to three years in a few cases, who have also shown persistent obliteration of varices embolized with a mixture of Gelfoam and Sotradecol.
REFERENCES 1. Baum S, Nusbaum M: The control of gastrointestinal hemorrhage by selective mesenteric arterial infusion of vasopressin. Radiology 98:497-505, Mar 1971 2. Bierman HR, Kelly KH, White LP, et al: Transhepatic venous catheterization and venography. JAMA 158: 1331-1334, 13 Aug 1955
3. Haverling M, Ovenfors C-O: Translumbar portography. A preliminary communication concerning a new method. Invest Radiol 3:376-381, Sep-Oct 1968 4. Kessler RE, Zimmon OS: Umbilical vein angiography. Radiology 87:841-844, Nov 1966 5. Lunderquist A: Personal communication 6. Lunderquist A, Vang J: Transhepatic catheterization and obliteration of the coronary vein in patients with portal hypertension and esophageal varices. N Engl J Med 291:646-649,26 Sep 1974 7. Man B, Kraus L, Pikielny S: Investigation of the portal circulation via the umbilical vein. Vasc Surg 8:193-202, May-Jun 1974 8. Nusbaum M, Baum S, Blakemore WS: Clinical experience with the diagnosis and management of gastrointestinal hemorrhage by selective mesenteric catheterization. Ann Surg 170:506-513, Sep 1969 9. Reuter SR, Chuang VP: Control of abdominal bleeding with autogenous embolized material. Radiologe 14:86-91, Feb 1974 10. Rosch J, Goldman ML, Dotter CT: Experimental catheter obstruction of the gastric coronary vein. Possible technique for percutaneous intravasculartamponade of the gastroesophageal varices. Invest Radiol 10:206-211, May-Jun 1975 11. Rousselot LM, Moreno AH, Panke WF: Studies on portal hypertension. IV. The clinical and physiopathologic significance of selfestablished (nonsurgical) portal systemic venous shunts. Ann Surg 150:384-410, Sep 1959 12. Steckel RJ, Rosch J, Ross G, et al: New developments in pharmacoangiography (and arterial pharmacotherapy) of the gastrointestinal tract. Invest Radiol 6: 199-211, May-Jun 1971 13. Viamonte M Jr, LePage J, Lunderquist A, et al: Selective catheterization of the portal vein and its tributaries. Preliminary report. Radiology 114:457-460, Feb 1975 14. Zeid SS, Felson B, Schiff L: Percutaneous splenoportal venography, with additional comments on transhepatic venography. Ann Intern Med 52:782-805, Apr 1960 Department of Radiology Veterans Administration Hospital Miami, Fla. 33140
New Techniques for Interruption of Gastroesophageal Venous Blood Flow 1 Raul Pereiras, M.D., Manuel Viamonte, Jr., M.D., Edward Russell, M.D., James LePage, M.D., Patrick White, M.D., and Duane Hutson, M.D. Enlarged gastroesophageal veins were successfully obliterated in 41 patients using embolization with modified autogenous clots and/or Gelfoam, balloon occlusion, iatrogenic perivenous hematoma, sclerosing agents (Sotradecol and Keflin), or a combination of these methods. Thirteen patients were actively bleeding when studied, and the site of bleeding was detected in 4. Surgical exploration of 16 patients and autopsy study of 5 showed persistent obliteration ranging between three weeks and seven months. No major complications requiring reparative surgery were encountered. Gelfoam soaked with Sotradecol is the preferred agent because it provides persistent obliteration of the emboli zed veins. Patients who are acutely bleeding or have done so previously are candidates for selective obliteration of the gastroesophageal veins. Embolism, therapeutic (Portal vein, therapeutic embolization, 9[57].129) • Esophagus, varices, 7 [ 1] .750 • Gastrointestinal tract, hemorrhage • Stomach, varices. 7[2].750
INDEX TERMS:
Radiology 124:313-323, August 1977
I UNDERQUIST and Vang recently reported on their use . . of selective obliteration of the coronary vein in 4 patients with portal hypertension and bleeding gastroesophageal varices (6). We have employed superselective catheterization of tributaries of the portal vein by the transhepatic approach and have applied this procedure to (a) the study and treatment of patients with portal hypertension, particularly those with actively bleeding gastroesophageal varices; (b) the differential diagnosis of posthepatitic jaundice; (c) the diagnosis of pancreatic disease; and (d) hemodynamic and physiological studies in patients with portal hypertension (13).
9 ml of blood) were used in 4 patients and a mixture of Gelfoam and Sotradecol in 38 (Figs. 1-11). Balloon occlusion of the coronary vein was employed in one case (Fig. 12). In 4 patients a perivenous hematoma obliterated the veins (Figs. 2 and 13). Intimal damage was induced prior to the use of thrombin. Keflin 2 was used in 2 patients and hypertonic glucose in 3. We were unable to catheterize the portal vein in 3 patients and the gastroesophageal veins in 6.
Technique (a) Gelfoam Administration: A surgical gelatin sponge {Gelfoam)3 is cut into small pieces and soaked in 3 % Sotradecol." The resulting gel-like suspension is injected slowly by hand using a 1-ml tuberculin syringe. Following 0.5 ml of the mixture, a small amount of contrast medium is administered under fluoroscopic control. From 2 to 6 ml of the mixture may be needed to obliterate the gastroesophageal veins successfully. It is unnecessary and potentially hazardous to obstruct the large veins close to the spleno-portal system, as embolic material may enter the system and obstruct intrahepatic branches of the portal vein and/or migrate into intestinal or pancreatic branches. (b) Catheterization: During the venous phase of indirect portography, the skin of the abdominal wall is marked at the points of anterior and lateral projection of the right branch of the portal vein (usually at the tip of the transverse process of T11 or T 12) and the patency of the vessel is established. We prefer to employ the highest possible point in the midaxillary line as the point of entry, paying careful
MATERIAL AND METHODS
Fifty-one studies were performed in 41 patients with hepatic cirrhosis and portal hypertension. Thirteen were bleeding actively at the time of the procedure. Nine patients were studied twice and one was studied three times. The coronary vein, short gastric veins, or both were superselectively catheterized in all cases. Before beginning the patient studies, horizontal, sagittal, and coronal sections from frozen cadavers were used to demonstrate the anatomical position of the bifurcation of the portal vein in the hilar portion of the liver. It is usually equidistant from the front and back of the body (though this varies with the physique of the patient and pathological conditions such as ascites and advanced liver disease) and generally projects ventrally and to the right of the spine at a distance equivalent to the anteroposterior diameter of L1. Modified autogenous blood clots (1 ml of Amicar plus
1 From the Departments of Radiology (R.P., P.W.) and Surgery (D.H.), Veterans Administration Hospital, the Department of Radiology, Jackson Memorial Hospital (M.V., E.R., J.L.P.), and the Department of Surgery, University of Miami School of Medicine (D.H.), Miami, Fla. Revised version accepted for publication in November 1976. 2 Eli Lilly & Co., Indianapolis, Ind. 3 Upjohn Co., Kalamazoo, Mich. 4 Elkins-Sinn, Inc., Cherry Hill, N.J. sjh
313
314
RAUL PEREIRAS AND OTHERS
August 1977
Fig. 1. CASE J. A and B. Embolization of the coronary vein with Gelfoam and perivenous infiltration of the short gastric veins. Superselective catheterization of the coronary vein reveals markedly enlarged gastroesophageal veins with intense opacification of the wall of the esophagus (arrows). Active bleeding was present, with oozing of contrast material. 0 tip of catheter.
=
attention to the farthest lateral inferior projection of the right costodiaphragmatic sulcus. After infiltrating the skin, subcutaneous plane, and intercostal space with a local anesthetic, a 21 gauge needle 5.1 cm (2 in.) long is used to penetrate the peritoneum and liver capsule while the patient stops breathing. A stab incision is made in the skin using a #11 blade, and the skin and subcutaneous tissues are separated with a straight forceps. A 19 gauge trocar'' 30-60 em long (average, 40 cm) with a plastic sleeve (o.d. 1.25 mm) is advanced close to the upper edge of the lower rib and directed parallel to the plane of the x-ray table. During apnea, the tip of the trocar is quickly directed toward the liver hilus (usually at the level of T11 or T12) under fluoroscopic control. The cannula and perforating needle are quickly removed and the patient is asked to breathe gently. Aspiration of blood followed by injection of contrast material will facilitate localization of a large 5 Becton-Dickinson Co., Rutherford N.J.; Johannah Medical Services, Inc., Minneapolis, Minn,
branch of the portal vein, after which a small J-shaped guide wire (o.d. 0.25 mm) is advanced through the catheter; it may take several passes before the portal vein is entered. Rotation of the guide wire will place it in the splenic, inferior mesenteric, or superior mesenteric vein as desired. If the coronary and/or short gastric veins are to be selectively catheterized, the end of the wire should be preshaped to facilitate entering these vessels (5). We have primarily used stainless steel Kifa guide wires. When the tip of the guide wire is in position, the plastic sleeve or other catheters are advanced over it. We have used catheters made of either polyethylene or Teflon (o.d. 1.35-1.58 mm). If catheters have to be exchanged, the patient is asked to breathe in and out and the portion of the guide wire between the liver and the chest wall is observed fluoroscopically to make sure that it is perpendicular to the chest wall, which is necessary to facilitate penetration with the catheter; technical difficulties may be encountered if the guide wire is obliquely directed toward the liver. Control of respiration is critical in this maneuver. Difficulties can
Vol. 124
NEW TECHNIQUES FOR INTERRUPTION OF GASTROESOPHAGEAL BLOOD FLOW
315
Diagnostic Radiology
Fig. 2. CASE I. A. After selective catheterization of the coronary vein (CV) with Gelfoam, Sotradecol, and thrombin, only small venules remain opacified in the esophagogastric region. B. Selective splenic venogram shows filling of the gastroesophageal veins (arrows) via the short gastric veins. Note that the coronary vein is completely occluded and not opacified from this injection. PV = portal vein; SV = splenic vein; 0 = tip of catheter. C. Scout radiograph shows an area of infiltration surrounding the tip of the guide wire. Although the gastric veins were selectively catheterized, the catheter could not be advanced into the smaller venules. A test injection showed extravasation of contrast material, and approximately 15 ml more was then injected to induce spasm. D. Postembolization venogram shows interruption of flow in the gastroesophageal region with nonfilling of the coronary vein and short gastric veins. PV = portal vein; SV = splenic vein.
also occur in patients with a small, scarred right lobe, such as those with advanced cirrhosis. Still under fluoroscopic control, the catheter is advanced to the level of the splenic hilus to obtain a control portogram after removing the gUide wire. This is followed by selective catheterization of all veins connecting the spleno-portal system with the fundus of the stomach and the distal esophagus. If the liver is hardened and the catheter cannot be advanced over the guide wire, as in patients with marked hepatic fibrosis or small portal vein branches, we suggest that coaxial catheterization be employed. Following completion of the study, we attempt to obliterate the juxtacapsular portion of the needle track to avoid
leakage of bile or blood into the peritoneal cavity. For this purpose, the catheter is withdrawn to a point about 2 cm from the surface of the liver and contrast material is injected slowly to ensure nonfilling of the hepatic veins, which if opacified would preclude embolization of the needle track. If the hepatic veins are visualized, the catheter should be withdrawn a few millimeters farther and the test injection repeated; if not, about 0.2 ml of Gelfoam is extruded with a guide wire. If hepatic venous opacification persists, Gelfoam might embolize into the lung; therefore we recommend that thrombin be used instead. We have studied patients with cholestatic jaundice and cirrhosis (most of whom also had ascites) without encountering major bleeding complications.
316
RAUL PEREIRAS AND OTHERS
Fig. 3. CASE 2. Embolization of the coronary vein with Gelfoam and temporary balloon occlusion of the azygos vein. A. Direct portogram shows hepatofugalflow with filling of the coronary (CV), inferior mesenteric (lMV), and umbilical veins (VV). PV = portal vein; SV = splenic vein. B. Selective catheterization of the coronary vein shows free communication with the azygos system (A). 0 = tip of catheter; SV = splenic vein. C. Temporary balloon occlusion (B) of the azygos vein. D. Following embolization of the coronary vein with Gelfoam, angiogram shows the portal (PV) and splenic veins (SV) with no filling of the gastroesophageal veins. Fig. 4. CASE 2. Late phase of the postocclusion portogram demonstrates hepatofugal flow with filling of the inferior mesenteric (lMV) and superior mesenteric veins (SMV). A few short gastric veins are filled (arrows); the patency of these veins may be the cause of subsequentbleeding. PV = portal vein; 0 = tip of catheter.
August 1977
Vol. 124
NEW TECHNIQUES FOR INTERRUPTION OF GASTROESOPHAGEAL BLOOD FLOW
317
Diagnostic Radiology
Fig. 5. CASE 3. Embolization of the short gastric and coronary veins with Gelfoam. A. Findings on the preliminary direct portogram are deceptive, showing the splenic (SV) and portal veins (PV) without demonstrating the gastroesophageal veins. Band C. Selective catheterization of the short gastric veins (SG). Note the partial drainage toward the left renal vein (arrow). D. Postembolization angiogram shows obstruction of the gastroesophageal veins (arrows). Note the short gastric (SG), splenic (SV), and tip of catheter. inferior mesenteric veins (/MV), not seen on the preembolization angiogram. 0 E. Direct portogram now reveals partial filling of the coronary vein (solid arrow) and some flow through one of the short gastric veins (broken arrow) toward the fundus of the stomach. PV = portal vein; SV = splenic vein.
=
RESULTS
Thirteen patients were studied while actively bleeding, but angiographic proof of the bleeding site was demonstrated in only 4. Bleeding was controlled by embolization in all 13 but recurred in 7: the cause of bleeding was incomplete vascular obliteration in 5, Mallory-Weiss syndrome in 1, and an esophageal ulcer in 1. Hemorrhage occurred one week after the procedure in 1 patient, three weeks afterward in 2, four months later in 1, and six months later in 1. Twenty-eight patients were studied while stable. In 9 we merely obliterated the coronary vein, as the patient was being considered for a distal spleno-renal shunt.
Of the 28 patients who are still alive (not the same group as the 28 just mentioned), 16 had surgery following the examination: 2 of them were lost to follow-up and the other 14 have not exhibited further bleeding. Of the 12 patients who did not have surgery, bleeding recurred in one of the 4 followed up for less than one month, none of the 3 studied for one to three months, one of the 2 followed up for three to six months, and none of the 3 studied for six to twelve months. In 4 patients, superselective catheterization of the coronary vein and short gastric veins was unsuccessful and inadvertent extravasation of contrast material around these vessels was noted. Surprisingly, portography revealed complete interruption of flow (Figs. 1 and 13),
318
RAUL PEREIRAS AND OTHERS
August 1977
Fig. 6. CASE 3. Despite poor visualization on the direct portogram, the coronary vein was selectively catheterized. A. Angiogram shows the coronary vein (CV) with an area of narrowing (solid arrows) which corresponds to spasm at the tip of the catheter. Note the filling of the gastroesophageal veins and the large varix (V) in the region of the distal esophagus (broken arrows). B. Postembolization angiogram demonstrates the coronary vein (Cv) and obliteration of the veins previously visualized in the distal esophagus. C. Final postembolization portogram reveals the splenic (SV) and portal veins (PV), but the gastroesophageal veins are not visible.
probably due to venous spasm and compression by a hematoma. Repeat study showed renewed permeability of these vessels, which were subsequently obliterated with a mixture of Gelfoam and Sotradecol. One of our patients had hepatic cirrhosis, portal hypertension, and partial thrombosis of the splenic and portal veins, with very slow blood flow in the gastroesophageal veins. Balloon occlusion of the coronary vein for 20 minutes following intimal trauma and injection of thrombin effectively interrupted flow (Fig. 12). In our patients, the mean pressure in the spleno-portal veins was 1 mm Hg higher than the wedged hepatic vein pressure. Following venous obliteration, the portal vein pressure is usually 2 mm Hg above preembolization levels. When other large porto-systemic collaterals were demonstrated, very little if any rise in portal vein pressure was noted after embolization. No significant complications were observed in our series. Almost all patients had a small subcapsular hematoma of the liver which resolved spontaneously. This was detected on the direct portogram obtained at the completion of embolization: the costal surface of the right lobe of the liver appeared concave and was occasionally separated from the abdominal wall. One patient had active intraperitoneal bleeding which required 1.4 I (3 pints) of blood but no surgical intervention: he demonstrated abnormal coagulation and was on a respirator. Another patient became anuric for four days following a procedure in which 400 ml of contrast medium was used. Renal function slowly returned to precatheterization levels about two weeks later. In one patient in whom only the coronary vein was embolized, massive bleeding occurred 24 hours later, causing death from bleeding esophageal varices. One patient had postcatheterization bleeding caused by inadvertent puncture of an intercostal artery, which was treated by suturing the vessel. Seven of the patients studied while actively bleeding and
2 of those studied while stable died. The cause of death was hepatoencephalopathy in 2, sepsis in 1, recurrent bleeding in 4, and unknown causes in 2. Four patients were lost to follow-up. DISCUSSION
The radiological treatment of gastrointestinal bleeding has included arterial approaches, such as selective, prolonged infusion of vasopressin (9) or a combination of epinephrine and propanolol (8), and injection of autogenous clots, Gelfoam, etc. into the arteries supplying the bleeding area (12), with varying degrees of success. In patients with portal hypertension, intra-arterial infusion of vasopressin has been used most often (1). The approach described here provides immediate and good results. Indirect portography beforehand is desirable but not necessary, and we have not used it in emergency situations. In order to produce effective, permanent obliteration of the gastroesophageal veins, a small catheter should first be advanced as close as possible to the small branches of the coronary and/or short gastric veins. Induction of intimal damage using a guide wire and injection of thrombin or a sclerosing agent as proposed by Lunderquist and Vang (6) may be an adequate method of obliterating veins. However, we believe that in high-flow situations, or when dealing with several veins giving off numerous branches, injection of embolic material into the main collecting trunk (such as the coronary vein) will be more effective. Assessment of the effectiveness of direct venous embolization requires more experience; however, we have documented persistent obliteration of the coronary and short gastric veins with Gelfoam and Sotradecol three weeks to twelve months after the examination and believe that Gelfoam embolization may give the best results. Esophagography is not a good means of differentiating patent from obliterated veins, as varices detected on the postembolization esophagogram may be totally obliterated.
Diagnostic Radiology
Fig. 7. CASE 4. Selective catheterization of three short gastric veins. A. Direct portogram demonstrates hepatofugal flow. PV = portal vein; SMV = superior mesenteric vein; SV splenic vein; C = collateral vein. X and Y represent two short gastric veins supplying the gastroesophageal region. The broken arrow indicates short gastric veins near the splenic hilus which were responsible for the area of extravasation shown in Figure 7, B. B. Late phase of the direct portogram. Note the esophageal varices ( V) and extravasationof contrast material along the greater curvature of the stomach (arrows). Fig. 8. CASE 4. A and B. Selective injection of a short gastric vein (Y) shows opacification of the gastroesophageal veins. Note the extravasation of contrast material into the stomach (broken arrows). C and D. Postembolization study. Arrows indicate the level of occlusion. SV == splenic vein.
=
319
320
RAUL PEREIRAS AND OTHERS
August 1977
Fig. 9. CASE 4. Selective injection of another short gastric vein (X). A. Preembolizationangiogram reveals prominent veins in the esophagogastric region. IMV inferior mesenteric vein; SV splenic vein. There is minimal extravasation of contrast material (broken.arrow). B. Postembolization angiogram shows nonfilling of the esophagogastric veins (solid arrows). Note the extravasation of contrast material (broken arrows) above and below the projection of the splenic vein (SV), which can be seen because of reflux into the splenic vein and subsequent passage into a third short gastric vein (Z). In the preliminary angiogram the area of extravasation was not marked. IMV = inferior mesenteric vein. C. Direct portogram obtained following selective embolization of two short gastric veins (X and Y)(solid arrows). Note the hepatofugal flow and the third short gastric vein (Z) with extravasation of contrast material (broken arrow) just below the projection of the splenic vein (SV). PV portal vein; SMV superior mesenteric vein; IMV inferior mesenteric vein. D. Late phase of the direct portogram shows marked extravasation of contrast material into the stomach (arrows). Note the opacification of one short gastric vein (Z). PV = portal vein. Fig. 10. CASE 4. ' A and B. Selective catheterization of the third short gastric vein (Z). Note the marked extravasation of contrast material into the stomach (arrows). C and D. Postembolization angiogram shows obliteration of the short gastric veins (solid arrows). Extravasation persists, but to a lesser degree (broken arrow).
=
=
=
If the gastroesophageal veins are not completely obliterated, variceal bleeding may recur. When patients are selected for a distal spleno-renal shunt, the coronary vein is ligatedat the time of surgery, while the short gastric
=
=
veins are left undisturbed to provide another outlet for decompression of the gastroesophageal region. Blood flows from the fundus of the stomach through the short gastric veins into the splenic veins and from them into the
Vol. 124
NEW TECHNIQUES FOR INTERRUPTION OF GASTROESOPHAGEAL BLOOD FLOW
Fig. 11. CASE 4. Final portogram. A. Early phase. Note the complete obstruction of the short gastric veins which were embolized (X, Y, Z) (straight arrows). There is hepatofugal flow with filling of the superior mesenteric (SMV) and inferior mesenteric veins (/MV) as well as a large collateral (C) connected to the splenic vein (SV). Filling of the coronary vein (CV) is seen for the first time (curved arrow), with minimal opacification of the esophagogastric veins. PV portal vein. B. Late phase. There is no extravasation of contrast material into the stomach. Note the opacification of the gonadal vein (GV) and inferior vena cava (IVC), suggesting spontaneous diversion of porto-systemic flow. The lack of evidence of active bleeding into the stomach and the considerable amount of contrast material used in performing the procedure explain why the coronary vein was not embolized and the study was discontinued. The patient died of hepatorenal syndrome eight days later. Numerous superficial gastric ulcers were present, but there was no evidence of recurrent bleeding at the esophagogastric junction.
=
321
Diagnostic Radiology
322
RAUL PEREIRAS AND OTHERS
August 1977
Fig. 12. CASE 5. A and B. Balloon occlusion of the coronary vein. Direct portogram revealed occlusion of the superior mesenteric and splenic veins (solid arrows) with hepatofugal flow (broken arrow) via the coronary vein (CV). The slow flow in the coronary vein motivated the use of balloon occlusion. PV = portal vein; 0 = tip of catheter; LPV = left portal vein. C. After inducing intimal damage with a guide wire and injecting thrombin through the catheter, a balloon catheter (B) was inflated in the coronary vein. D. After 15 minutes of balloon occlusion, total interruption of flow in the coronary vein was observed on a control angiogram. PV = portal vein; CV = coronary vein.
left renal vein. In such cases we have obliterated only the coronary vein. We recommend that surgery be performed shortly after transhepatic obliteration because of the possibility of recurrent bleeding from the patent short gastric veins. Patientstreated only with embolization shouldbe studied again four to eight weeks later. If adequate spontaneous porto-systemic communications such as spleno-renal or mesocaval vessels are lacking, other short gastric veins may enlarge and promote further variceal bleeding. In such cases, embolization should be repeated until no gastroesophageal veins are demonstrated angiographically. A potential hazard of this technique is passage of clots or Gelfoam into the azygos system and from there into the pulmonary circulation via the right side of the heart. Temporary balloon occlusion of the azygos vein may prevent this. However, pulmonary perfusion studies made 24 hours after the procedure in 12 patients studied without temporary balloon occlusion of the azygos vein and in 5 autopsies failed to demonstrate pulmonary embolism.
Contraindications include a bleeding tendency and the presence of a hypervascular mass in the path of the catheter. Fever, hypersensitivity to contrast agents, ascites, anemia, and the need for a respirator are considered risk factors but not contraindications. Vital signs must be monitored closely for at least eight hours after the procedure. In cadaver and cast-corrosion studies of patients with cirrhosis of the liver and portal hypertension, we have observed intense vascular engorgement of the spleen contrasted with a reduced hepatic venous bed; these findings coupled with the friability of this organ contraindicate transsplenic catheterization of the portal vein. We have not used the cumbersome transjugular (10) and translumbar (3) approaches. Catheterization of the umbilical portal vein (4, 7) requires extraperitoneal dissection, and we have used it in only a few patients. Transhepatic portography is not a new technique (2, 14). However, selective catheterization of the portal vein and its tributaries for diagnosis and treatment is a new devel-
Vol. 124
NEW TECHNIQUES FOR INTERRUPTION OF GASTROESOPHAGEAL BLOOD FLOW
323
Diagnostic Radiology
Fig. 13. CASE 6. Perivenous infiltration of the coronary vein in a patient with thrombosis of an earlier mesocaval shunt. A. Direct portogram showed hepatofugal flow with filling of the superior mesenteric (SMV), inferior mesenteric (lMV), and coronary veins (CV). Contrast material injected by hand in an attempt to advance the coronary vein catheter into smaller veins appeared to extravasate; nevertheless, approximately 15 ml was injected. PV portal vein; SV splenic vein; arrows indicate the direction of venous flow. B. A second injection was carried out with the tip of the catheter in the splenic vein (0). Note the extravasated contrast material in the gastroesophageal region (broken arrows); flow is spontaneously diverted into the left renal vein (LRV) and from there to the inferior vena cava (lVC) through large retroperitoneal veins which empty into the left spermatic and left adrenal veins (solid arrows). C. A third injection was carried out with the tip of the catheter (0) at the junction of the splenic (SV) and portal veins (PV). Broken arrows indicate interruption of flow in the coronary vein secondary to perivascular extravasation of contrast material; solid arrows indicate the direction inferior mesenteric vein. of venous flow. SMV = superior mesenteric vein; IMV
=
=
=
opment (6, 13). When the platelet count is below 40,000 or the prothrombin level is 7 over control, the umbilical approach appears to be safest. The transjugular technique has been proposed in such situations and also when ascites is present (10). In our patients with ascites (80 % ), we used the transhepatic approach only with hesitation. None required surgical intervention following the procedure or medical treatment for bleeding.
SUMMARY Bleeding from ruptured esophageal varices is the most common cause of death in patients with cirrhosis and portal hypertension. We believe that transhepatic catheterization and obliteration of the gastroesophageal veins with Gelfoam and Sotradecol is a safe and effective method of halting or preventing bleeding in such cases. ADDENDUM: Since this paper was submitted, we have studied more than 80 patients, with follow-up for up to three years in a few cases, who have also shown persistent obliteration of varices embolized with a mixture of Gelfoam and Sotradecol.
REFERENCES 1. Baum S, Nusbaum M: The control of gastrointestinal hemorrhage by selective mesenteric arterial infusion of vasopressin. Radiology 98:497-505, Mar 1971 2. Bierman HR, Kelly KH, White LP, et al: Transhepatic venous catheterization and venography. JAMA 158: 1331-1334, 13 Aug 1955
3. Haverling M, Ovenfors C-O: Translumbar portography. A preliminary communication concerning a new method. Invest Radiol 3:376-381, Sep-Oct 1968 4. Kessler RE, Zimmon OS: Umbilical vein angiography. Radiology 87:841-844, Nov 1966 5. Lunderquist A: Personal communication 6. Lunderquist A, Vang J: Transhepatic catheterization and obliteration of the coronary vein in patients with portal hypertension and esophageal varices. N Engl J Med 291:646-649,26 Sep 1974 7. Man B, Kraus L, Pikielny S: Investigation of the portal circulation via the umbilical vein. Vasc Surg 8:193-202, May-Jun 1974 8. Nusbaum M, Baum S, Blakemore WS: Clinical experience with the diagnosis and management of gastrointestinal hemorrhage by selective mesenteric catheterization. Ann Surg 170:506-513, Sep 1969 9. Reuter SR, Chuang VP: Control of abdominal bleeding with autogenous embolized material. Radiologe 14:86-91, Feb 1974 10. Rosch J, Goldman ML, Dotter CT: Experimental catheter obstruction of the gastric coronary vein. Possible technique for percutaneous intravasculartamponade of the gastroesophageal varices. Invest Radiol 10:206-211, May-Jun 1975 11. Rousselot LM, Moreno AH, Panke WF: Studies on portal hypertension. IV. The clinical and physiopathologic significance of selfestablished (nonsurgical) portal systemic venous shunts. Ann Surg 150:384-410, Sep 1959 12. Steckel RJ, Rosch J, Ross G, et al: New developments in pharmacoangiography (and arterial pharmacotherapy) of the gastrointestinal tract. Invest Radiol 6: 199-211, May-Jun 1971 13. Viamonte M Jr, LePage J, Lunderquist A, et al: Selective catheterization of the portal vein and its tributaries. Preliminary report. Radiology 114:457-460, Feb 1975 14. Zeid SS, Felson B, Schiff L: Percutaneous splenoportal venography, with additional comments on transhepatic venography. Ann Intern Med 52:782-805, Apr 1960 Department of Radiology Veterans Administration Hospital Miami, Fla. 33140
