Hemobilia after Percutaneous Transhepatic Biliary Drainage: Treatment with Transcatheter Embolotherapy

Hemobilia after Percutaneous Transhepatic Biliary Drainage: Treatment with Transcatheter Embolotherapyl Scott J. Savader, MD Scott o. Trerotola, MD Dimitri S. Merine, MD Anthony C. Venbrux, MD Floyd A. Osterman, MD

Index terms: Bile ducts, hemorrhage, 76.46 • Bile ducts, percutaneous drainage, 76.1263 • Hepatic arteries, therapeutic blockade, 95.129 • Interventional procedures, complications, 76.46 JVIR 1992; 3:345-352 Abbreviation: iary drainage

PBD

=

percutaneous bil-

1 From the Russell H. Morgan Department of Radiology and Radiological Science, Blalock 545-CVDL, Johns Hopkins Medical Institutions, 600 N Wolfe St, Baltimore, MD 21205. Received September 23, 1991; accepted November 21. Address reprint requests to S.J.S.


SCVIR, 1992

Thirteen of 333 patients who underwent percutaneous biliary drainage (PBD) developed severe hemobilia. Hepatic arteriography successfully demonstrated the source of hemorrhage in all 13 patients. Lesions included hepatic artery pseudoaneurysm in nine, hepatic artery-bile duct fistulas in four, and a hepatic artery-portal vein fistula in one patient. Hemobilia occurred from 1 day to 1.8 years (mean, 100 days) following catheter placement. Embolization agents used included Hilal embolization microcoils, occluding spring emboli, cyanoacrylate, detachable balloons, and gelatin sponge pledgets. A single agent was used in eight cases (62%), multiple agents were used in four cases (31 %), and in one case (7%), spontaneous thrombosis of the pseudoaneurysm occurred during catheter manipulation. In five patients, the source of the hemorrhage could only be demonstrated following removal of the biliary catheter(s) over guide wire(s). Initial embolization was successful in stopping hemobilia in 12 patients. One patient required repeat embolization after 4 months. Postembolization complications included hepatic abscess formation in two patients and a sterile hepatic infarct in one patient. This series indicates that transcatheter embolotherapy is an effective method for the treatment for severe hemobilia.

PERCUTANEOUS biliary catheters are frequently used for drainage and stent placement in the treatment of patients with both benign and malignant biliary obstruction (1-5). Because percutaneous biliary drainage (PBD) is an invasive procedure, potential complications can occur, including cholangitis (13%-47%), sepsis (3.9%-8.0%), pancreatitis (0-10%), hemobilia (2.6%-9.6%), and death (.06%-5.6%) (1,6-11). Hemobilia, even if severe, is rarely a cause of death in these patients (5,7,9,12); however, prompt angiographic evaluation and transcatheter embolotherapy is warranted to avert a lifethreatening situation, particularly in the debilitated individual. Embolization of hepatic artery lesions has been performed with a variety of agents (12,13). In this report, we discuss our experiences with transcatheter embolotherapy in 13 patients who

developed hemobilia following placement ofPBD catheters.

PATIENTS AND METHODS Between July 1989 and January 1991,333 patients at our institution underwent PBD with use of standard techniques. Of these, 169 had benign and 164 had malignant biliary disease. Thirteen patients (eight men, five women), whose ages ranged from 46 to 82 years (mean, 67 years), who developed hemobilia as a direct result of PBD presented for percutaneous transcatheter embolotherapy. Clinical data on the study group is summarized in the Table. Patients with asymptomatic hemobilia (ie, nonpulsatile bleeding through the PBD catheter with stable blood pressure and hematocrit) at presentation were initially evaluated

345

346 • Journal of Vascular and Interventional Radiology May 1992

Summary of Patient Data Patient No./ Age (y)/Sex

Diagnosis

Stent Type

Time to Bleed * (d)

Source of Bleeding

Total Units of Blood t

1/80/M

Klatskin tumor

2-16-F HS

27

RHAPA

0

Treatment 0.2 mL cyanoacrylate, four M-coils

2/77/F

Periampullary carcinoma Pancreatic carcinoma Pancreatic carcinoma

10-F Cope

8

RHAPA

I

I-mm balloon

8.3-F Ring

28

RHAPA

2

10-FVTC

7

RHA-BD fistula

15

Eight3-mm M-coils Spontaneous thrombosis

12-FVTC

9

RHAPA

5

Six 3-mm M-coils

8-FKaufman

1

RHA-PV fistula

0

1-mm balloon 0.1 mL cyanoacrylate 0.4 mL cyanoacrylate one 3-mm coil, two 5-mm coils, one 8-mm coil, gelatin sponge

3/49/M 4/67/F

5/51/F 6/66/M

Klatskin tumor Sclerosing cholangitis Klatskin tumor Sclerosing cholangitis

2-16-F HS, 18-F HS 2-16-F HS

663

RHAPA

6

63

RHAPA

42

9/82/M

Gallbladder carcinoma

8.3-F Ring

89

RHAPA

2

10/66/M

Gastric carcinoma

8-FVTC

23

RHA-BD fistula

5

11/46/M

Pancreatic carcinoma

8.3-F Ring

6

RHAPA

4

12/61/F

Pancreatic carcinoma Klatskin tumor

2-10-FVTC

RHA-BD fistula 2RHAPA

5

7/67/M 8/78/M

13/76/M

8-FKaufman

365 8

2

One 8-mm coil, three 5-mm coils, two 3-mm coils I-mm balloon (initial); four 5-mm coils (repeat embolization) Four3-mm M-coils, two 3-mm coils, two 5-mm coils Two 3-mm M-coils Cyanoacrylate, gelatin sponge, 10 3-mm coils

Results Bleeding stopped; developed sterile hepatic infarct 4 days after embolization Bleeding stopped Bleeding stopped Bleeding stopped spontaneously, exsanguinated during heparin therapy for DVT Bleeding stopped Bleeding stopped; died of primary disease Bleeding stopped; died of cancer Bleeding stopped; developed hepatic abscess 19 days after embolization; no hemobilia in 17 months Bleeding stopped; died of cancer Bleeding stopped but recurred within 4 months, requiring coil embolization Bleeding stopped; died of cancer Bleeding stopped Bleeding stopped; developed hepatic abscess 16 days after embolization

Note.-BD = biliary duct, Cope = Cope biliary catheter (Cook, Bloomington, Ind), DVT = deep venous thrombosis, HS = HeyerSchulte soft Silastic catheter (Mentor, Goleta, Calif), Kaufman = Kaufman multiple-side-hole straight drain (Cook), M-coil = Hilal microembolization coil (Cook), PA = pseudoaneurysm, PV = portal vein, RHA = right hepatic artery, Ring = Ring biliary catheter (Cook), VTC = Percuflex biliary system (Medi-tech/Boston Scientific, Watertown, Mass). * Period between stent placement and presentation with hemobilia. t Total units of whole blood transfused prior to successful embolization.

Savader et al • 347 Volume 3 Number 2

a h ~ Figure 1. Patient 9. Right hepatic artery pseudoaneurysm in a 82-year-old man with gallbladder carcinoma (Ring biliary catheter removed 7 weeks prior to presentation). (a) Celiac axis arteriogram demonstrates a 5-mm pseudoaneurysm (arrow) arising from the right hepatic artery just distal to the origin of the left hepatic artery. No extravasation is noted at this time. (b) Selective catheterization with contrast material injection adjacent to the pseudoaneurysm (small arrow) demonstrates marked extravasation with rapid filling of the common bile duct (large arrows). (c) Mter coil embolization, the proper hepatic artery is occluded (arrow) proximal to the right and left hepatic artery. Hepatic infarction did not occur.

with cholangiography to ensure that the catheters were properly positioned and that no side holes were within the hepatic parenchyma. Malpositioned catheters were repositioned. Tube size was increased if there were no side holes within the parenchyma but blood was tracking back along the catheter and tube tract. Patients with symptomatic hemobilia (ie, hypotension, tachycardia, falling hematocrit, or pulsatile blood loss through the PBD catheter) at presentation and with properly positioned catheters were believed to have an arterial source of hemorrhage and underwent immediate arteriography including celiac axis, selective hepatic, and superior mesenteric arteriography. Portal vein patency was confirmed by reviewing the portal venous phase of the superior mesenteric arteriogram either from this study or from a recent arteriographic evaluation. Catheterization was performed from a transfemoral approach. Diagnostic studies were generally performed with use of a 5-F femoral sheath. Balloon embolotherapy required up to a 10-F sheath. Magnification and oblique views combined with digital subtrac-

tion arteriography were routinely used during selective hepatic artery injections to define the injured branch vessel. Selective catheterization for embolotherapy of the hepatic artery or its branches was performed with use of either a 4.9-F nontapered catheter, a 5- or 6.5-F cobra catheter, or a 6.5-F nontapered Simmons shaped catheter. A Tracker 18 catheter (Target Therapeutics, San Jose, Calif) was directed through one of the above catheters forming a coaxial system in nine cases. In one case in which bleeding was from a right hepatic artery pseudoaneurysm, a transhepatic biliary drainage catheter tract was embolized directly through an 8-F sheath. Embolization was planned for each patient based on the angiographic findings and in such a fashion as to preserve maximum hepatic parenchyma. Selective or subselective catheterization was attempted in each patient based on arterial anatomy and lesion location. The embolic agent(s) used for embolotherapy varied and the choice of agent(s) was made by the individual radiologist. Patients received ampicillin and gentamycin intravenously prior to the

initiation of the procedure and for the next 7-10 days. Six embolizations were performed with use of occluding spring emboli (Cook), Hilal embolization microcoils (Cook), or a combination of the two. Occluding spring emboli were delivered by using a 5-F cobra-type catheter in two cases (Fig 1). Hilal microcoils were delivered at the site of arterial injury in three cases with a Tracker 18 catheter (Fig 2). In one case in which both types of coils were required, a Simmons I-Tracker 18 coaxial system was required for coil delivery. Three embolizations were performed with I-mm detachable balloons (Mini-Balloon; Bard-Parker, Lincoln Park, NJ). Selective hepatic arterial access was obtained prior to "floating" the balloon on its 2-F catheter from the guiding catheter (4.9-F nontapered catheter in two cases, 6.5-F nontapered cobra catheter in one case) to the site of arterial injury (Fig 3). The balloons were detached only after test injections demonstrated the balloons had been properly positioned across the neck of the arterial injury. In four patients, embolization was accomplished with a combination of agents including cya-

348 • Journal of Vascular and Interventional Radiology May 1992

L

~

~

Figure 2. Patient 12. Right hepatic artery-bile duct fistula in a 61-year-old woman with pancreatic carcinoma (bilateral IO-F VTC biliary catheters). (a) Celiac axis arteriogram obtained after the PBD catheters were removed over guide wires demonstrates extravasation of contrast material from a right hepatic artery branch into the common bile duct (arrows). (b) A Tracker-I8 catheter (arrow) was advanced through the Simmons catheter to the lesion for delivery of microcoils. (c) Mter embolization, arteriogram shows multiple microcoils (arrows) occluding the right hepatic artery-common bile duct fistula. Note the lack of contrast material extravasation into the common bile duct.

noacrylate (0.2 mL) and microcoils (patient 0; cyanoacrylate (0.4 mL), gelatin sponge pledgets (Gelfoam; Upjohn, Kalamazoo, Mich) and occluding spring emboli (patients 8 and 13); and cyanoacrylate (0.1 mL) alone (patient 7). In one patient, no embolization was performed. RESULTS During the 19-month period of this study, 333 patients underwent percutaneous biliary drainage. Fourteen episodes of symptomatic hemobilia occurred in 13 patients (3.9%), and 26 episodes of asymptomatic hemobilia occured in 26 patients (7.8%). In the asymptomatic group, all cases of hemobilia resolved either following tube repositioning or tube upsizing. In the symptomatic group, 14 hepatic artery lesions were identified at arteriography: nine hepatic artery pseudoaneurysms (ie, fistula with angiographically demonstrated extravasation), four hepatic artery-bile duct fistulas, and one hepatic arteryportal vein fistula. Eleven cases (79%) of hemobiIia occurred in pa-

tients with percutaneously placed biliary drainage catheters ranging in size from 8 to 12 F. Three cases (21%) of hemobilia occurred in patients whose PBD catheter had been upsized during surgery to a larger Silastic Heyer-Schulte biliary stent ranging in size from 16 to 18 F. Hemobilia required replacement of 0-42 units (mean, 5.6 units) of whole blood prior to embolization of the arterial lesion. Eleven of 13 patients (85%) had malignant disease (pancreatic carcinoma in four, Klatskin tumor in four, periampullary carcinoma in one, gallbladder carcinoma in one, and gastric carcinoma in one) and two patients (15%) had benign disease (sclerosing cholangitis in both). Patients presented with hemobilia from 1 day to 1.8 years (mean, 100 days) following initial PBD catheter placement. Angiography was successful in identifying the source of hemorrhage during the initial study in 11 patients. In five of these patients (42%), the arterial lesion could only be defined during a second arterial contrast material injection following removal of the biliary catheter(s) over a

guide wire (Fig 4). Three patients required repeat arteriography prior to successful embolization. In patient 13, a second arteriogram was required (with the PBD catheter removed over a guide wire), which helped identify two pseudoaneurysms 11 days following the initial study. Initial failure was most likely secondary to failure to remove the PBD catheter prior to the study. One pseudoaneurysm was successfully embolized with cyanoacrylate. The second pseudoaneurysm could not effectively be embolized transarterially without loss of significant hepatic parenchyma. Therefore, with use of Gelfoam pledgets and multiple 3-mm occluding spring emboli, the percutaneous tube tract crossing the arterial injury was embolized through an 8-F sheath as it was withdrawn. In patient 8, empiric Gelfoam embolization of the left hepatic artery was performed after the third angiogram failed to demonstrate a site of vascular injury despite three episodes of hemobilia solely from the left PBD catheter. During the fourth arteriographic study, however, both PBD catheters were removed over

Savader et al • 349 Volume 3 Number 2

r

a.

b. c. Figure 3. Patient 2. Right hepatic artery pseudoaneurysm in a 77-year-old woman with periampullary carcinoma (IO-F Cope biliary drainage catheter). (a) Digital subtraction radiograph of the right hepatic artery branches demonstrates a 1.5-cm pseudoaneurysm (arrow) adjacent to the PBD catheter (arrowheads). (b) Digital subtraction radiograph demonstrates a I-mm detachable balloon (arrow) in the branch vessel supplying the pseudoaneurysm. (c) Mter detachment of the I-mm balloon, digital subtraction arteriogram demonstrates nonfilling of the pseudoaneurysm. Note the excellent preservation of arterial supply peripheral to the site of embolization due to collateral circulation.

L

~

Figure 4. Patient 4. Right hepatic artery-bile duct fistula in a 67-year-old woman with pancreatic carcinoma (IO-F VTC biliary catheter). (a) Celiac axis arteriogram demonstrates relatively normal-appearing intrahepatic vessels and no extravasation. Note the position of the PBD catheter (arrows). (b) Celiac axis arteriogram following removal of the PBD over a guide wire demonstrates contrast material extravasation (arrows) into the PBD tract via a right hepatic artery-bile duct fistula. Spontaneous thrombosis of this fistula occurred during attempted catheterization of the branch vessel.

350 • Journal of Vascular and Interventional Radiology May 1992

guide wires and a right hepatic artery pseudoaneurysm was identified, which was subsequently embolized with cyanoacrylate, Gelfoam, and occluding spring emboli. Patient 10 required repeat embolization 4 months later because the detachable balloon migrated distally, reopening the hepatic artery-bile duct fistula. Occluding spring emboli were successfully used to terminate the hemobilia. Patient 4 experienced spontaneous thrombosis of the arterial branch supplying the hepatic artery-bile duct fistula during catheter manipulations within the vessel. Further problems were encountered obtaining adequate catheter purchase without sacrificing an inordinate amount of liver; thus, no further embolization was performed in this patient. Transcatheter embolotherapy was successful in controlling hemobilia in 12 patients. In patient 4, attempted embolization resulted in spontaneous thrombosis of the branch vessel supplying the pseudoaneurysm prior to delivery of the embolic agent, presumably from catheter manipulation. This patient's condition was stable for 2 weeks, but she was then readmitted with deep venous thrombosis. Heparin was administered, but she subsequently exsanguinated before our department was consulted. Multiple embolic agents including steel coils (occluding spring emboli and Hilal embolization microcoils), cyanoacrylate, detachable balloons, and Gelfoam were used both individually (62%) and in combination (31%). Embolization coils used alone were successful in stopping hemobilia in six lesions. Embolization coils were used in combination with other agents included cyanoacrylate (one lesion), and Gelfoam and cyanoacrylate (two lesions). Detachable balloons were used alone successfully in two of three lesions and cyanoacrylate was used alone successfully in two lesions. One patient underwent empiric left hepatic artery embolization with Gelfoam. Mter technically successful embolization of a demonstrated vascular lesion, hemobilia recurred only in patient 10 (who underwent a suc-

cessful repeat embolization with coils). Follow-up for this group ofpatients ranged from 1.5 to 17 months. Eleven patients have subsequently undergone routine biliary tube changes one to 14 times (mean, four) either up to the time of their death or until the present. One patient has not required a routine tube change since the embolization, and one patient died of her malignancy prior to a tube change. To date, eight patients are alive, four patients have died of their malignancies, and one patient in whom no embolization was performed, exsanguinated after heparin therapy was begun. Of the 13 patients who underwent PBD, eight had dilated biliary systems and five had nondilated biliary systems. Surgeons at our institution commonly use percutaneous biliary catheters to aid in definition of anatomic landmarks and in surgical reconstruction for patients with both benign and malignant biliary disease. Thus patients with nonobstructive hepatobiliary disease may often have one PBD catheter placed prior to surgery (1,2,4,13). Eight of244 patients (3.3%) with dilated biliary systems developed arterial lesions while five of 89 patients (5.6%) with nondilated biliary systems developed hepatic arteriallesions as a direct result of PBD. Two patients developed intrahepatic abscesses 16 and 19 days (mean, 14.5 days), respectively, following arterial embolization. Both patients were given antibiotic therapy (ampicillin, gentamycin) prior to the embolization and for the period until their abscess was diagnosed. Embolization agents in these two cases included cyanoacrylate (one patient) and Gelfoam pledgets (one patient). In both cases, percutaneous transhepatic drainage of the abscess was performed under fluoroscopic guidance. Culture and sensitivities of the abscess contents showed that in both patients, organisms were present (group D streptococci [enterococci], Morganella morganii, and coagulasenegative staphylococci in patient 8

and Klebsiella pneumoniae in patient 13) that were resistant to ampicillin and gentamycin therapy. Patients were switched to appropriate antibiotic therapy until the abscess resolved. Neither patient required surgical evacuation of the abscess. Patient 1 developed a sterile hepatic infarct after embolization with cyanoacrylate and microcoils. Prophylactic antibiotic therapy was administered but catheter evacuation was not necessary. There were no other complications.

DISCUSSION Intrahepatic vascular injuries following PBD can occur in a significant percentage of patients. In a study by Hoevels and Nilsson, 33% of patients demonstrated arterial lesions at hepatic angiography following PBD (14). Only 19% of these lesions, however, required follow-up therapy, suggesting that the majority of these injuries are self-limiting. Monden et al demonstrated a 13.8% prevalence of hemobilia in patients (6.4% with severe hemobilia), with 19.1% of those studied arteriographically demonstrating vascular lesions (15). More recently, Mitchell et al demonstrated 5% frequency of severe hemobilia after PBD of 397 patients (12). Our study compares favorably, demonstrating an overall prevalence of severe hemobilia after PBD of 3.9% in 333 patients. Our 7.8% prevalence of asymptomatic hemobilia also compares favorably with other studies in the literature (15). In all of our cases, asymptomatic hemobilia was effectively treated with either tube repositioning or tube upsizing. It is noteworthy that of the patients with arterial lesions after PBD, 5.62% had nondilated biliary systems versus 3.28% of patients with dilated biliary systems, yielding a ratio of 1.7 times increased injury in the nondilated group. Since the portal triad contains the hepatic artery, bile duct, and portal vein all close together, one could hypothesize that the risk of arterial injury should decrease as the

Savader et al • 351 Volume 3 Number 2

size of the adjacent bile duct increases. The Fischer exact test was applied to this data, and while the results of our study may suggest that a slightly greater risk of arterial injury is incurred in the patient with a nondilated biliary system (p = .22), overall this does not appear statistically significant. Further studies do seem warranted, however, and we continue to collect data at this time. Arteriographic evaluation should include selective hepatic artery studies in addition to evaluation of the celiac axis. We have noted that the celiac axis arteriogram may fail to help identify active extravasation, whereas a selective hepatic arteriogram will show significant bleeding (patients 9 and 13). In cases of massive hemobilia, embolization may have to be performed without a superior mesenteric arteriogram with portal venous phase. In this situation, the risk of hepatic infarction is greatly increased if there is portal venous compromise (ie, tumor encasement). Whenever possible, evaluation of the portal venous system should preceed embolization, or one should have knowledge of its patency (ie, recent portal venogram or Doppler study). Arteriographic identification of vascular lesions in these patients may be significantly enhanced when the study is performed without the biliary tubes in place. In five cases (42%), the vascular injury was only visible following removal of the biliary catheters over a guide wire (Fig 4). Thus it seems that in a significant percentage of cases, "tube tamponade" may be responsible for an initial false-negative arteriogram. In all patients, initial arteriographic evaluation should be performed with the tubes in place to avoid any undue or life-threatening hemorrhage. However, should arteriographic evaluation fail to demonstrate a site of arterial injury, the biliary tubes should be removed over guide wires prior to repeat contrast material injection. In addition, patients with bilateral tubes should have both tubes removed prior to repeat evaluation, as the pre-

sentation of hemobilia may not always occur through the biliary tube responsible for the arterial injury. This was clearly demonstrated with patient 8 during multiple episodes of hemobilia. A variety of embolic agents have been used successfully in the hepatic circulation both in this study and in others (12,13,16). Embolization coils were used in our study to embolize six lesions. Embolization coils are inexpensive, may be delivered quickly with relative ease, are permanent, and preserve hepatic parenchyma. Their disadvantage lies in that superselective catheterization is required. The coils must be delivered precisely beginning at a sight just distal to the arterial injury and extending across the lesion so that the neck of the pseudoaneurysm or fistula is bridged. Delivery of coils too proximal (subselective embolization of arterial branches not feeding the lesion) can result both in continued hemobilia from collateral circulation and infarction of large segments of the hepatic parenchyma, particularly in patients with concomitant liver disease or compromised portal venous circulation. We found the Tracker catheter system for superselective coil embolization extremely valuable in that peripherally located lesions could be treated without sacrificing significant hepatic parenchyma. Detachable balloons were used in three cases. Though not inexpensive or easily delivered, they are advantageous for other reasons. Permanent occlusion with excellent preservation of hepatic parenchyma has been demonstrated by Mitchell et al and others (12,13). In addition, they can be deflated and repositioned prior to final detachment should contrast material test injections demonstrate suboptimal positioning. Detachable balloons, however, do require a certain level of skill and expertise to use. Delivery often requires more than simply "floating" the balloon out to the lesion, and, once deployed, they may migrate distally (12). Cyanoacrylate was used alone to treat one lesion and in combination

with other agents to treat two additionallesions. Cyanoacrylate is mixed with iophendylate (Pantopaque; Lafayette Pharmacol, Lafayette, Ind) in different ratios to produce a liquid embolic agent with a predetermined "polymerization time" (17). This agent, however, is difficult to handle, timing of polymerization is complex (and must be exact), and distal embolization results in peripheral arterial occlusion, potentially leading to hepatic infarction and abscess formation (patient 13). Cyanoacrylate is most likely not an optimal agent for embolizations within the liver. Gelfoam is inexpensive and easy to use but it is not permanent and even large pledgets may be carried into peripheral arteries by rapid hepatic arterial flow, thus resulting in hepatic infarction or insufficiency (patient 8) (14). Gelfoam was used in our study with coils in two cases. Gelfoam can be a very useful agent when it is placed between embolization coils to help promote stasis and thrombosis or when placed directly within the pseudoaneurysm from a transarterial or a transhepatic approach (12,16). Gelfoam can be placed in the hepatic artery branches "unprotected" (without coils on either side); however, it may embolize proximally or distally, resulting in unintentional vessel occlusion with potentially disastrous results. If Gelfoam is to be used in this fashion, care should be given to demonstrate portal vein patency in the hepatic segment to be embolized and proper pre- and postprocedural antibiotic therapy is a necessity. Two patients in this study who received pre- and postembolization antibiotic therapy developed postembolization hepatic abscesses. The embolic agents included cyanoacrylate and Gelfoam. It has been demonstrated that intrahepatic abscess formation can be a common complication in patients with long-term PBD catheters, and the mortality rate can be high (66%) (18). In addition, studies have shown that biliary obstruction is a predisposition to hepatic infarction following arterial em-

352 • Journal of Vascular and Interventional Radiology May 1992

bolization (19). Dilated biliary ducts are thought to compress adjacent portal venous structures, thus rendering the hepatic parenchyma dependent on arterial blood supply. Consequently, arterial embolization in these patients devascularizes the hepatic parenchyma, resulting in necrosis. This has even been shown to occur in patients with obstructions who have already undergone PBD and decompression (14). This pattern of arterial-dominant circulation in conjunction with a biliary tract colonized by bacteria and fungus (from the presence of the PBD catheter) is the optimal setting for post-embolization abscess formation. In our two patients who developed hepatic abscesses despite antibiotic therapy, we can only theorize that a combination of selective biliary tract colonization with bacteria resistant to antibiotics typically appropriate for the biliary tract, and altered circulatory hemodynamics were responsible for this unfortunate complication. Mitchell et al achieved excellent results after embolization with detachable balloons (12). On the basis of our results, and supported by the work of others, we believe that for hepatic arterial lesions, embolization coils and detachable balloons are the safest embolization agents, offering both a high level of control in terms of delivery, and allowing for the greatest degree of preservation of peripheral circulation (12,13). Geifoam pledgets are safe to use, particularly when "sandwiched" between embolization coils. Cyanoacrylate is not the agent of choice in these instances. In cases in which the arterial lesion is located very peripherally (patient 13), one runs the risk of hepatic infarction with any agent. Perhaps in this setting, it may be best to attempt to treat the arterial lesion through selective embolization of the tube tract at its point of communication with the injured vessel or via direct percutaneous puncture of the pseudoaneurysm under fluoroscopic control (16). Surgery has long been the treatment for hemobilia. Studies have shown, however, that surgery is often

a major procedure, it may require several operations to eventually control hemobilia, and the complication rate may be high (16,20-22). In addition, surgical ligation of the hepatic artery may not be effective in controlling hemobilia secondary to intrahepatic collateralization, collateral circulation from the hepatoduodenal ligament or capsular branches arising from the phrenic arteries (16). Transcatheter embolotherapy provides an effective alternative to surgery as verified by our series of patients and follow-up. If initial arteriographic evaluation is negative, the study can be repeated (without the biliary tubes in place) until the arterial lesion is identified. This allows for selective embolization of the appropriate branch vessel versus the surgical alternative of hepatic artery ligation, thus preserving maximum hepatic parenchyma. In addition, transcatheter embolotherapy can usually be performed in debilitated patients who are poor surgical risks. Acknowledgment:

The authors express their sincere appreciation to Debbie Offenbacker for her expertise in the preparation of this manuscript.

References 1.

2.

3.

4.

5.

6.

Savader SJ, Venbrux AC, Robbins KV, GittelsohnAM, Osterman FA. Pancreatic response to percutaneous biliary drainage: a prospective study. Radiology 1991; 178:343-346. Zuidema GD, Cameron JL, Sitzman JV, et al. Percutaneous transhepatic management of complex biliary problems. Ann Surg 1983; 197:584-593. Harrington DP, Barth KH, Maddrey WC, Kaufman SL, Cameron JL. Percutaneously placed biliary stents in the management of malignant biliary obstruction. Dig Dis Sci 1979; 24:849857. Gallacher DJ, Kadir S, Kaufman SL, et al. Nonoperative management of benign postoperative biliary strictures. Radiology 1985; 156:625-529. Gunther RW, Schild H, Thelen M. Review article: percutaneous transhepatic biliary drainage: experience with 311 procedures. Cardiovasc Intervent Radio11988; 11:65-71. Mueller PR, vanSonnenberg E, Ferrucci JT Jr. Percutaneous biliary drainage: technical and catheter related problems in 200 procedures. AJR 1982; 138: 17-23.

7. Yee ACN, Ho C. Complications of percutaneous biliary drainage: benign vs malignant diseases. AJR 1987; 148: 1207-1209. 8. Clark RA, Mitchell SE, Colley DP, Alexander E. Percutaneous catheter biliary decompression. AJR 1981; 137: 503-509. 9. Carrasco IT, Zernoza J, Bechtel WJ. Malignant biliary obstruction: complications of percutaneous biliary drainage. Radiology 1984; 152:343-346. 10. OleagaJA, RingEJ. Interventional biliary radiology. In: Felson B, ed. Seminars in roentgenology. Vol 16. New York: Grune & Stratton, 1981; 116-134. 11. Kadir S, Baassiri A, Barth KH, Kaufman SL, Cameron JL, White RI Jr. Percutaneous drainage in the management of biliary sepsis. AJR 1982; 138:25-29. 12. Mitchell SE, Shuman LS, Kaufman SL, et al. Biliary catheter drainage complicated by hemobilia: treatment by balloon embolotherapy. Radiology 1985; 157:645-652. 13. Cameron JL, Gayler BW, Herlong HF, Mackdrey WC. Sclerosing cholangitis: biliary reconstruction with silastic transhepatic stents. Surgery 1983; 94: 324-330. 14. Hoevels J, Nilsson U. Intrahepatic vascular lesions following nonsurgical percutaneous transhepatic bile duct intubation. Gastrointest Radio11980; 5:127-135. 15. Monden M, Okamura J, Kobayashi N, et al. Hemobilia after percutaneous transhepatic biliary drainage. Arch Surg 1980; 115:161-164. 16. Uflacker R, Mourno GS, Piske RL, Souza VC, Lima S. Hemobilia: transcatheter occlusive therapy and longterm follow-up. Cardiovasc Intervent Radio11989; 12:136-141. 17. Widlus DM, Lammert GK, Brant A, et al. In vivo evaluation of pantopaquecyanoacrylate mixtures. Radiology (in press). 18. Pennington L, Kaufman S, Cameron JL. Intrahepatic abscess as a complication of long-term percutaneous internal biliary drainage. Surgery 1982; 92: 642-645. 19. Doppman JL, Girton G, Vermess M. The risk of hepatic artery embolization in the presence of obstructive jaundice. Radiology 1982; 143:37-43. 20. Steichen FM, Scheiner NM. Traumatic intrahepatic hemobilia. Arch Surg 1966; 92:838-847. 21. Enge IP, Knutrud 0, Normann T. Central rupture of the liver with traumatic hemobilia. Br J Radio11968; 41: 789-791. 22. Katz MC, Meng CH. Angiographic evaluation of traumatic intrahepatic pseudoaneurysm and hemobilia. Radiology 1970; 94:95-99.