Selective arterial embolization for life threatening hemobilia after transjugular intrahepatic portosystemic shunt placement

174

Letters to the Editor

Selective arterial embolization for life threatening hemobilia after transjugular intrahepatic portosystemic shunt placement To the Editor: Hemobilia is a well-known complication of percutaneous transhepatic interventions, such as liver biopsy or percutaneous transhepatic cholangiography. It has also been reported after transjugular intrahepatic portosystemic shunt (TIPS) placement in up to 5% of procedures, but in most cases it is self-limited [1] and does not require speci®c treatment. We herein report the management of a patient who developed persistent hemobilia after the placement of TIPS, leading to severe deterioration in his clinical condition. A 47-year-old man with alcoholic cirrhosis was admitted to our hospital in March 1998 because of a variceal rebleeding episode after failure of pharmacological and endoscopic treatment. After the initial control with urgent sclerotherapy, a TIPS was placed and, because of a deteriorated liver function, evaluation for liver transplantation was initiated. TIPS was successfully carried out without immediate complications. After TIPS the portal pressure gradient decreased from 15 to 10 mmHg and he was discharged from hospital 13 days after the procedure free of bleeding. On the 19th day after TIPS, the patient was admitted again because of severe right upper quadrant abdominal pain. Physical examination revealed tachycardia, hypotension, jaundice and tenderness on right upper quadrant. Hemoglobin dropped from 10.4 to 6.8 mg/dl. Three hours after admission he had melena and hematemesis. Endoscopy showed small varices and blood oozing from the ampulla. An ultrasonography showed intra- and extrahepatic bile duct dilation with echogenic material inside and a patent shunt. An urgent hepatic arteriography was performed, failing to disclose an arterio-biliary or arterioportal ®stula. A high pressure oppaci®cation of the porto-hepatic tract, occluding the stent with a balloon catheter, was performed without demonstrating any ®stula. Nine units of packed red cells were transfused, and 24 h later the patient was stable with no further signs of bleeding. Two days later the patient complained again of severe abdominal pain. He was sweaty and hypotense, and a bruit could be listened over the liver surface. He had melena and hematemesis and 1 h later he underwent a second negative hepatic arteriography. The patient continued with intermittent hemobilia, requiring the transfusion of 3±4 units of packed red blood cells per day. On day 27 a covered stent was coaxially placed in an attempt to seal a possible undetected biliary-shunt ®stula, but bleeding persisted and a third negative angiography was performed. However, a small distal branch of the right hepatic artery close to the stent

was empirically embolized with ivalon contour particles without success. On day 32 after TIPS, patient was still bleeding, and a new arteriography disclosed an arterioportal ®stula that originated from the superior branch of the right hepatic artery that had not been seen in previous angiograms (Fig. 1). No opaci®cation of the biliary tree was observed. A supraselective embolization of the arterial branch with three coils was performed. Hemobilia ceased and no further transfusion was required, ALT and AST remained unchanged and 6 days after embolization the patient was discharged. One year after the episode, duplex±Doppler ultrasound and angiographyc studies show that the TIPS is patent. The patient is free of ascites but with severe liver disfunction and occasional self limited encephalopathy. He is still waiting for liver transplantation. Hemobilia is the result of a porto-biliary or arteriobiliary ®stula. In the absence of portal hypertension, the normal pressure in the common bile duct overwhelms the pressure in the portal vein [2]. In that case, a porto-biliary ®stula results in bilemia rather than hemobilia, with recurrent bacteriemia, marked increase in serum bilirrubin and hemolysis [1]. In the cirrhotic patient, portal pressure may be higher than the pressure of the biliary tract even after TIPS decompression leading to the development of both hemobilia and recurrent bacteriemia. [3]. Hemobilia may be massive if an arterio-biliary ®stula is created [4]. Arterio-portal ®stula has also been reported in patients with hemobilia after other transhepatic percutaneous procedures, such as liver biopsy or transhepatic biliary drainage [5]. The success in controlling hemobilia after angiographic embolization suggests an intermittent three-way ®stula communication with the biliary tract. Development of an arterio-portal ®stula has been reported after TIPS placement, resulting in deteriorating liver function [6], but, to our knowledge, our case is the ®rst describing an arterioportal ®stula in the setting of a TIPS induced hemobilia. Diagnostic approach to hemobilia after TIPS placement should include both portography and hepatic artery angiography. However, as shown in our case, bleeding may be intermittent, hence making the diagnosis dif®cult. In our case three hepatic artery angiographies and two portographies performed during apparent active bleeding failed to detect the ®stula. It is possible that a sudden rise in biliary tract pressure during bleeding may promote the intermittent closure of the ®stula. The thrombogenic properties of bile [7] may also play a role in the spontaneous closure of the ®stula. In cases of portobiliary ®stula located in either the stent tract or the intrahepatic portal vein, a covered stent may be placed transjugulary to seal the ®stula [8]. Performance of

Letters to the Editor

175

Fig. 1. Selective hepatic arteriography showing the partial ®lling of the portal venous axis (arrows).

this procedure is easy and safe. For this reason, this was our ®rst line treatment in the absence of a topographical diagnosis of the ®stula. Hemobilia resulting from arterio-biliary or arterio-portal ®stula is most commonly treated with arterial embolization in the absence of a TIPS [5]. However, patients who have undergone a TIPS have most or all of their portal ¯ow diverted to the systemic circulation, and hepatic perfusion is highly dependent on hepatic artery ¯ow. This is clearly demonstrated by reports showing extensive hepatic infarctions and accelerated liver failure after hepatic artery injuries during the placement of a TIPS [9]. Consequently, the risk of hepatic artery embolization is only acceptable if it is performed in a superselective way. This makes it mandatory to reach an exact topographical diagnosis, because blind superselective embolization has a greater probability of being unsuccessful. In our case an arterioportal ®stula, identi®ed after three negative arteriographies, was embolized with coils. This proved to be extremely effective. The hemobilia ceased and an improvement in the liver function tests and in the clinical condition of the patient was observed. In summary, this case shows that hemobilia may be a life

threatening complication after TIPS placement. In the case of transient ®stulas, perseverance in performing arteriography while actively bleeding is the only way to establish a proper topographical diagnosis. Hepatic artery embolization may be a safe and effective therapy for this condition if performed in a targeted superselective way. Juan GonzaÂlez-Abraldes 1, Eduardo Moitinho 1, Juan C. Garcõ a-PagaÂn 1, Angels Escorsell 1, Juan M. SalmeroÂn 1, Rosa Gilabert 1, Maribel Real 2, Xavier Muntanya 2, Jaime Bosch 1 1 Liver Unit, Institut de Malalties Digestives, Hospital ClõÂnic, Vilarroel 170, 08036 Barcelona, Spain 2Interventional Radiology Unit, Hospital ClõÂnic, Institut de Investigacions BiomeÁiques August Pi i Sunyer, University of Barcelona, Barcelona, Spain

References [1] Rossle M, Siegerstetter V, Huber M, Ochs A. The ®rst decade of the transjugular intrahepatic portosystemic shunt (TIPS): state of the art. Liver 1998;18(2):73±89. [2] Verhille MS, Munoz SJ. Acute biliary-vascular ®stula following needle aspiration of the liver. Gastroenterology 1991;101(6):1731±1733. [3] Willner IR, El-Sakr R, Werkman RF, Taylor WZ, Riely CA. A ®stula

176

Letters to the Editor

from the portal vein to the bile duct: an unusual complication of transjugular intrahepatic portosystemic shunt. Am J Gastroenterol 1998;93(10):1952±1955. [4] Menzel J, Vestring T, Foerster EC, Haag K, Roessle M, Domschke W. Arterio-biliary ®stula after transjugular intrahepatic portosystemic shunt: a life-threatening complication of the new technique for therapy of portal hypertension. Z Gastroenterol 1995;33(5):255±259. [5] Hidalgo F, Narvaez JA, Rene M, Dominguez J, Sancho C, Montanya X. Treatment of hemobilia with selective hepatic artery embolization. J Vasc Interv Radiol 1995;6(5):793±798. [6] Sedat J, Padovani B, Chanalet S. Arterioportal ®stula after transjugular intrahepatic portosystemic shunt placement. Am J Roentgenol 1995;164(1):259.

[7] Teng GJ, Bettmann MA, Hoopes PJ, Wagner RJ, Park BH, Yang L, Baxter BR. Transjugular intrahepatic portosystemic shunt: effect of bile leak on smooth muscle cell proliferation. Radiology 1998;208(3):799±805. [8] Spahr L, Sahai A, Lahaie R, Dufresne MP, Bui BT, Dagenais M, Fenyves D, Pomier-Layrargues G. Transient healing of TIPS-induced biliovenous ®stula by PTFE- covered stent graft. Dig Dis Sci 1996;41(11):2229±2232. [9] Haskal ZJ, Pentecost MJ, Rubin RA. Hepatic arterial injury after transjugular intrahepatic portosystemic shunt placement: report of two cases. Radiology 1993;188(1):85±88.

Improvement of magnetic resonance spectroscopic abnormalities but not pallidal hyperintensity followed amelioration of hepatic encephalopathy after occlusion of a large spleno±renal shunt To the Editor: Proton-magnetic resonance (MR) studies of the brain in patients with cirrhosis show a characteristic pallidal T1-hyperintensity and, using spectroscopy, an increase in glutamine and a decrease in myo-inositol and choline [1± 3]. The interpretation of these abnormalities and their relationship with hepatic encephalopathy is controversial. In order to better understand the signi®cance of MR abnormalities, we would like to report the evolution of pallidal hyperintensity and magnetic resonance spectroscopic abnormalities after occlusion of a large spleno±renal shunt in a patient with cirrhosis and chronic relapsing hepatic encephalopathy. A 67-year-old female with cirrhosis due to hepatitis C and a history of 10 years of chronic relapsing hepatic encephalopathy was submitted to an angiographic study that found a large spleno±renal shunt. She had neither gastrointestinal varices nor ascites and showed a relatively preserved liver function (albumin 22 g/l, prothrombin time INR 1.2, total bilirubin 2.1 mg/dl, platelets 83 000 £ 10 9/l). During all these years, she followed a protein-restricted diet, branched chain amino acids supplementation and continuous lactulose therapy. It was decided to occlude the large spleno±renal shunt by means of tungsten coils. After shunt occlusion the patient followed the same anti-encephalopathy treatment. Before shunt occlusion she was admitted to the hospital on seven occasions for hepatic encephalopathy; most of the episodes were spontaneous or precipitated by respiratory or urinary infections. In addition she had minor episodes of encephalopathy that were treated at home with enemas of lactulose. At the time of shunt occlusion, she exhibited mild action tremor and mild exacerbation of deep tendon re¯exes; there were no clinical signs of parkinsonism. After shunt occlusion, she developed only one episode of hepatic encephalopathy, 2 months after embolization. This episode was mild and precipitated by diuretics that were

prescribed because of ascites. One year after shunt occlusion, the patient did not exhibit changes on the standard neurological examination but she did exhibit signi®cant improvements on several neuropsychological tests (Fig. 1). The improvement on neuropsychological function was clinically relevant, as demonstrated by improvements in her daily life autonomy: the Barthel's scale improved from 85 to 100% (the Barthel's scale ranges from 100%: full autonomy to 0%: full dependence) and the Everyday Omission Questionnaire [4] (ful®lled by her husband) decreased from 236 to 121 (this scale ranges from 48 to 288; mean ^ SD ˆ 81 ^ 18 in 40 healthy subjects). Two years after shunt occlusion she died because of a respiratory infection. In order to assess the effects of shunt embolization, she

Fig. 1. Neuropsychological tests [13] at baseline and 1 year after shunt occlusion. Results are expressed as T scores calculated from normal values of general population. T scores range from 0 to 100, being 50 the mean value for a speci®c age group. A decrease in 10 points (from 50 to 40) represents a score that is 1 SD below the mean value. The following neuropsychological tests were grouped according to the main cognitive domain. Attention: Oral Symbol Digit Test (1), Trail Making Test part A (2), Stroop Test - colour-words (3). Memory: Auditory Verbal Learning-trial 1 (4), Auditory Verbal Learning-total learning (5), Auditory Verbal Learning-recall (6). Motor: Grooved Pegboard-dominant (7), Grooved Pegboard-non-dominant (8). Language: COWA Test (9). Executive function: Hooper Visual Test (10).