bigger diameter. In this situation, one may use a semicom- pliant balloon, the diameter of which increases slightly with increase in inflation pressure. Of course, coils for emboliza- tion should be ready before stent deployment. Another potential alternative is use of a self-expandable covered stent, such as the Wallgraft (Boston Scientific, Min- neapolis, MN) or other covered Nitinol stents. They still exert considerable pressure and the risk of rupture still persists. In conclusion, one has to be meticulous and extremely careful if a covered stent is to be used for treatment of acute carotid blow-out syndrome. Reference 1. Macdonald S, Gan J, McKay AJ, et al. Endovascular treatment of acute carotid blow-out syndrome. J Vasc Interv Radiol 2000; 11:1184 –1188. Fatal Pulmonary Embolus after TIPS Revision From: Eric K. Hoffer, MD John J. Borsa, MD Robert D. Bloch, MD Arthur B. Fontaine, MD Department of Radiology Harborview Medical Center 325 9th Avenue Box 359728 Seattle, WA 98104 Editor: The transjugular intrahepatic portosystemic shunt (TIPS) is an effective treatment for variceal hemorrhage and ascites secondary to portal hypertension. However, active surveil- lance and shunt revisions are required to overcome the poor 1-year primary patency rates (20%– 69%) and achieve sec- ondary 1-year patency rates of 90% (1,2). Revision entails balloon dilation and possible additional stent placement at sites of stenosis or occlusion. Thrombolysis is infrequently used to recanalize the shunt tract, and mechanical (balloon) maceration of clot has not been associated with significant morbidity (1–3). In this letter, we report a case of an oc- cluded shunt tract, lengthened by revisions, in which bal- loon dilation resulted in a fatal pulmonary embolus. The patient was a 61-year-old woman with a history of alcohol-related liver disease. At another hospital, she pre- sented with upper gastrointestinal bleeding and esopha- goduodenoscopy demonstrated varices, gastropathy, and a duodenal ulcer. She was treated with TIPS creation. A year later, she returned with upper gastrointestinal bleeding and was transferred to our hospital. Portography showed that the two Wallstents (Boston Scientific/Medi-tech, Natick, MA) that comprised the original TIPS were separated and had a stenosis between them. After a 10-mm  68-mm Wallstent was deployed across the two existing stents, the entire tract was dilated to 10 mm, with a resultant reduction in gradient from 13 mm Hg to 9 mm Hg. There was no further bleeding. A year later, routine follow-up US revealed occlusion of her shunt. Transjugular portography confirmed the occlu- sion and, after placement of three additional Wallstents and dilation to 10 mm, the gradient was reduced to 2 mm Hg (Fig 1). There was no further filling of the coronary varix and inferior mesenteric vein flow was antegrade. Five months later, the patient returned with a 2-week history of progressive, massive ascites and US showed an occluded shunt. Via right femoral access, the shunt tract was easily crossed with a guide wire, and contrast study con- firmed the occlusion (Fig 2). Because of the length of the occluded tract and the likelihood of an acute component of the thrombus, a 5-F pulse-spray catheter (Angiodynamics, Queensbury, NY) was advanced over the guide wire and 250,000 U urokinase (Abbokinase; Abbott Laboratories, North Chicago, IL) in 10 mL saline was administered over a 15-minute period, lacing the length of the occluded stents. The shunt was then dilated with a 10-mm-diameter, 4-cm angioplasty balloon (Ultra-thin; Boston Scientific/Medi- tech), beginning at the portal end with sequential inflations at 3-cm intervals back to the hepatic vein. After inflation at the junction of the hepatic vein and inferior vena cava, the patient reported acute pain and became unresponsive and cyanotic. Her oxygen count fell rapidly despite immediate use of the nonrebreathing mask with 100% oxygen. She became bradycardic and went into electromechanical dissociation. She died despite full advanced cardiopulmo- nary life support arrest protocol. At autopsy, there was a large, acute thromboembolus in the right main pulmonary artery and multiple smaller emboli in peripheral pulmonary vessels. There were residual clots in the TIPS tract and no deep venous thrombi identified in the lower extremities. Although some investigators use thrombolytic drugs in the setting of acute thrombosis the day after shunt creation, recanalization of an occluded shunt is most often performed by balloon dilation with or without stent placement (1,2). LaBerge and colleagues (1) routinely performed repeat stent placement and balloon dilation in occluded shunts, except in one case with extensive intra- and extrahepatic portal vein thrombosis, in which fibrinolytic therapy was used. Evi- dence that longer shunt tracts may be declotted without Figure 1. Transfemoral direct portography after the occluded shunt was treated with additional stents, with further extension of the shunt into the splenic vein (arrow). 896 • Letters to the Editor July 2001 JVIR