Liver Transplantation Complicated by Malpositioned Transjugular Intrahe- patic Portosystemic Shunts1 Mark W. Wilson, MD PURPOSE: To report the surgical problems encountered during or- Roy L. Gordon, MD thotopic liver transplantation as a result of a malpositioned trans- Jeanne M. LaBerge, MD jugular intrahepatic portosystemic shunt (TIPS). Robert K. Kerlan, MD PATIENTS AND METHODS: Three patients are described in whom Paul M. Radosevich, MD TIPS stents were malpositioned in the following locations: extend- John P. Roberts, MD ing into the main portal vein, extending into the suprahepatic infe- Ernest J. Ring, MD rior vena cava, and extending into the right atrium.

RESULTS: Malpositioning of TIPS stents altered and prolonged the operation in all of these patients by interfering with cross-clamping

Index terms: Liver, transplantation, at the usual vascular sites during liver transplantation. Incorpora- 761.4613 Shunts, portosystemic, 957,453,957,458

tion of the stents into the vascular wall prevented transcatheter re- trieval and increased the difficulty of intraoperative removal.

JVIR 1995; 6:695-699 CONCLUSION: Awareness of hepatic vascular anatomy is necessary Abbreviations: IVC = inferior vena in avoiding stent malpositioning. If malpositioning is identified, cava, TIPS = transjugular intrahepatic transcatheter approaches may be helpful in repositioning the stent. portosystemic shunt Otherwise, the transplant surgery team must be made aware of the

problem for proper surgical planning prior to liver transplantation.

Pmmwrs undergoing transjugular plicated liver transplantation sur- intrahepatic portosystemic shunt (TIPS) creation may require ortho- topic liver transplantation a t a later date. Indeed, a major application of CASE REPORTS the TIPS procedure is as a bridge to hepatic transplantation. Stenosis of Case 1.-A 31-year-old man un- TIPS is a common delayed complica- derwent TIPS for recurrent variceal tion, often necessitating revision bleeding secondary to cirrhosis from with deployment of additional metal- hepatitis C. The method of TIPS lic stents to improve shunt patency. placement followed previously de- Inadvertent malpositioning of the scribed techniques (1). The shunt expandable metallic stents used in was formed with a single 68-mm- the primary TIPS procedure or any long Wallstent device (Schneider revisions may complicate the surgi- USA, Minneapolis, Minn) dilated to cal transplantation procedure. If the 10 mm. Recurrence of variceal bleed- upper margin of the stent extends ing necessitated reexamination with into the right atrium or the inferior venography and modification of the vena cava (IVC) above the hepatic TIPS via the transjugular route on veins, cross-clamping of the IVC two separate occasions. The first

From the Department of Radiology, Uni- may trap the stent. Similarly, a modification was dilation and stent versity of 505 stent that extends too far into the placement across a hepatic vein ste- Parnassus Ave, San Francisco, CA 94143- 0628, Received January 6, 1995; revision portal vein may interfere with portal nosis. Two 68-mm-long Wallstents requested February 27; revision received vein cross-clamping and the portal were placed across the hepatic vein April 19; accepted April 25. Address re- vein anastomosis. stenosis and dilated to 10 mm, re- print requests to R.L.G. We report on three cases in which sulting in a decrease in the portosys- o SCVIR, 1995 malpositioned metallic stents com- temic pressure gradient from 21 mm

695

696 Journal of Vascular and Interventional Radiology September-October 1995

Figure 1. Case 1. A 31-year-old man b underwent TIPS for recurrent variceal bleeding secondary to cirrhosis from hepatitis C. (a) Follow-up transjugular portogram demonstrates a right hepatic vein stenosis (the initial shunt was formed with a single 68-mm long Wall- stent dilated to 10 mm). Arrows denote the ends of the stent. (b) Portogram ob- tained after extension of the upper end of the original shunt with two additional 68-mm Wallstents placed across the he- patic vein stenosis and dilated to 10 mm, resulting in a decrease in the portosys- temic pressure gradient from 21 mm Hg to 12 mm Hg. Arrows denote the ends of the stents. ( c ) Transjugular portogram i obtained 70 days later demonstrates re- , , current hepatic vein stenosis (arrow) and venous collaterals (arrowheads). The pressure gradient was 35 mm Hg. (d) TIPS revision with placement of a 30-mm Palmaz stent a t the upper mar- gin of the shunt, and deployment of two 42-mm Wallstents at the lower margin of the shunt, all of which were dilated to L L 10 mm. The lowermost Wallstent ex-

-1 a. d.

tends deep into the main portal vein. Ar- rows denote the hepatic and portal ends of the stents.

Hg to 12 mm Hg (Fig la , lb). The second modification 70 days later for recurrent hepatic vein stenosis and suboptimal extension of the inferior

; -1 end of the stent into the portal vein included placement of a 30-mm-long Palmaz stent (Johnson & Johnson Interventional Systems, Warren, NJ) at the upper margin of the shunt and deployment of two 42- mm-long Wallstents at the lower margin of the shunt, all of which were dilated to 10 mm (Fig lc, Id).

r The ~ressure cradient across the r L shuLt decreasld from 35 mm Hg to 22 mm Hg.

Eleven days after the second modification, the patient underwent liver transplantation. At surgery, a part of the lowermost Wallstent was noted to extend into the main portal vein to the level of the pancreas. To preserve this portion of the native portal vein, surgical separation and retrieval of the stent from the vessel was required. The portal anastomo- sis could then be performed and the

operation completed without compli- cation.

Case 2.-A 26-year-old woman underwent TIPS for recurrent vari- ceal bleeding secondary to autoim- mune hepatitis. The shunt was formed with a single 68-mm-long Wallstent dilated to 10 mm (Fig 2a). The initial TIPS lowered her porto- systemic pressure gradient from 30

mm Hg to 10 mm Hg. At routine fol- low-up 7 months later, hepatic vein stenosis was demonstrated with ul- trasonography and transjugular por- tography (Fig 2b). The hepatic vein stenosis was treated with insertion of a Wallstent dilated to 8 mm (Fig 2c). The pressure gradient returned to 10 mm Hg.

Three months later, the patient

Wilson et a1 697

Volume 6 Number 5

Figure 2. Case 2. A 26-year-old woman underwent TIPS for recurrent variceal bleeding secondary to autoimmune hepatitis. (a) Initial TIPS result; shunt was cre- ated with a single 68-mm Wallstent dilated to 10 mm. Arrows denote the ends of the stent. Note metallic coil embolization of venous collaterals (arrowheads). (b) Porto- gram obtained 7 months later demonstrates hepatic vein stenosis (arrow) and recur- rent venous collaterals (arrowheads). (c) TIPS revision with addition of a single 68- mm Wallstent dilated to 8 mm. The newly added upper stent extends into the supra- hepatic IVC. The final portosystemic pressure gradient was 10 mm Hg. Arrows denote the ends of the stents.

underwent the previously planned liver transplantation via a bilateral subcostal incision. At surgery, the uppermost Wallstent was discovered extending into the suprahepatic IVC, with incorporation into the caval wall. A clamp was applied across the suprahepatic cava and

the Wallstent was divided, allowing removal of the recipient liver. The residual portion of the stent was dis- sected from the caval wall and re- moved. The operation was then com- pleted without complication.

Case 3.-A 47-year-old man un- derwent TIPS a t an outside institu- tion approximately 4 weeks earlier for recurrent variceal bleeding sec- ondary to cirrhosis from hepatitis C. A single Wallstent was used that was 96 mm long by 10 mm in diam- eter. The upper margin of the stent was noted to extend into the right atrium on echocardiography. At- tempts were made to retrieve the stent by using a snare catheter from transjugular and transfemoral ap- proaches under fluoroscopic guid- ance (Fig 3). These were unsuccess- ful, as the stent seemed firmly incor- porated into wall of the hepatic vein

or the hepatic parenchymal tract. The patient underwent liver

transplantation the next day. The operation was performed by way of a midline incision with a median sternotomy. On palpation, the stent was felt within the right atrium. Af- ter the preparations for hepatectomy were completed, an incision was made through a purse-string suture in the right atrial appendage, allow- ing the surgeon to insert his finger into the atrium and push the stent down to the IVC below the dia- phragm. The cava was then cross- clamped above the stent, and the stent was removed along with the recipient liver in standard fashion. The remainder of the operation pro- gressed without complication.

DISCUSSION

Patients with advanced liver dis- ease may undergo a TIPS procedure a s therapy for variceal bleeding or intractable ascites. Liver transplan- tation is frequently an option for these patients a t a variable time af- ter TIPS. In the cases presented, this time interval has varied from 11 days to 3 months.

Liver transplantation is usually performed through a generous ab- dominal exposure, which allows vas- cular clamping of the short segment of suprahepatic IVC as i t passes through the diaphragm into the chest. This aspect of the operation may become complicated if the up- per portion of a metallic TIPS stent extends into the suprahepatic IVC, as i t will be trapped by cross-clamp- ing the very short segment of the IVC a t this level.

Nakamura and Tsuzuki described in detail the anatomy of the hepatic veins and suprahepatic IVC (2). They found that the average length of the IVC between the diaphragm and the upper margin of the right hepatic vein measures 0.7 cm k 0.3. I t is easy to appreciate the difficulty inherent in cross-clamping the su- prahepatic IVC under standard con- ditions, which could be complicated

698 Journal of Vascular and Interventional Radiology September-October 1995

considerably by even short lengths of metallic stent extending into the su- prahepatic IVC. The same study identified an average minimum caval length of 0.9 cm f 0.2 between the diaphragm and the inferior margin of the right atrium. This results in an average minimum caval length of 1.6 cm f 0.4 between the upper margin of the right hepatic vein and the infe- rior margin of the right atrium. An awareness of the short distances is key to preventing intraatrial stent deployment during the TIPS proce- dure.

The confluence of the right hepatic vein and the IVC can be difficult to accurately identify fluoroscopically due to respiratory and cardiac mo- tion. The Wallstent is only moder- ately radiopaque and is often poorly seen in obese patients or patients with ascites. The shortening of the stent, and hence the final location of a. b. the stent's end, is not always easy to predict. Figure 3. Case 3. A 47-year-old man underwent TIPS for recurrent variceal bleed-

ing secondary to cirrhosis from hepatitis C. (a) Transfemoral cavogram demon- In the third case an at- strates the superior margin of the Wallstent (arrow) extending into the right atrium. tempt was made pull the (b) Attempted retrieval of Wallstent from right atrium with use of a snare catheter.

engaging it with a nitinol snare The stent could only be partially bent away from the right atrium (arrows), suggest- from transjugular and transfemoral ing incorporation into the hepatic parenchymal tract or the hepatic vein wall. approaches. This technique has been used successfully by other authors in the early postdeployment period (3,4). The stent was implanted ap- on the stent struts immediately af- As experience with TIPS has in- proximately 4 weeks earlier and was ter intravascular deployment; fibrin creased, it has become clear that lu- probably incorporated into the caval deposition stimulates endothelial minal narrowing may occur within wall, explaining the lack of success in cell growth, forming a continuous the metallic stent itself and/or removing the stent in the case pre- layer covering the stent in 1-2 within the adjacent hepatic vein, sented here. In retrospect, it is pos- weeks; a neointima, composed of hy- leading to an increase in the porto- sible that the atrial end of the stent perplastic smooth muscle cell and systemic pressure gradient (11). could have been snared from a femo- inflammatory cells, forms between This may lead to placement of long ral approach prior to surgery. It then the stent struts and the endothe- or coaxial stents extending toward could have been bent and pulled lium; the endothelium-covered in- the junction of the hepatic vein with down out of the atrium with use of tima results in smoothing of the lu- the IVC or the main portal vein. the snare, thereby allowing cross- minal surface and incorporation of Thus, patients who require second- clamping of the cava without having the stent into the vessel wall; the ary stent placement after narrowing to open the atrium. Figure 3b shows neointimal thickness varies, but or occlusion of the originally placed that the stent was capable of being seems to stabilize by 6 months. shunt may be a t greater risk of stent bent away from the atrium, but the Woodle et a1 reported difficulty in malpositioning complicating subse- value of this maneuver was not ap- removing incorporated metallic quent liver transplantation. preciated a t the time. stents from blood vessels (10). It The availability of metallic stents

The histologic results of endovas- may be inferred that malpositioned of varying lengths (Wallstents now cular metallic stent deployment have stents are more likely to be retriev- range in size from 10 mm in diam- been investigated in animal models able in the acute or early postde- eter by 42 mm or 96 mm in length to and explanted livers (5-9). The fol- ployment period rather than a t a 12 mm in diameter by 40 mm or 90 lowing series of events have been later stage when incorporation may mm in length) provides the interven- noted: fibrin thrombus is deposited have occurred. tionalist with additional flexibility.

Wilson et a1 699

Volume 6 Number 5

The choice of which stent length to use is often subjective, as is the choice of where exactly the stent is deployed. When the optimum site of stent deployment is chosen, it is nec- essary to know how much a stent of a given length will shorten when completely deployed. Since no for- mulas exist relating degree of short- ening to optimal deployment site, success in this endeavor is probably a function of the interventionalist's experience coupled with knowledge of the essential venous anatomy of the liver. However, in some circum- stances, use of external radiopaque markers (eg, a small hemostat clamped to the overlying sterile drape) may be helpful to denote the safe limits for stent deployment based on contrast material iniec- tions into the pertinent vascular structures. Parallax must be ac- counted for when this or similar techniques are used. If currently available Wallstents were judged to be too long in a given case, a shorter Palmaz stent may be considered in- stead. If the Wallstent is judged to be too long a t either end after it has been deployed, overdilating the stent with larger balloons may bring about further shortening.

It is important to identify pa- tients with malpositioning of TIPS stents and bring this to the atten- tion of the surgical team prior to transplantation. This allows the proper surgical approach to be planned in advance (eg, a median sternotomy), if preoperative reposi- tioning by means of transvenous manipulation is unsuccessful. There is the potential for injury to the sur- geon attempting to manually remove a metallic stent due to its pointed ends. This is of importance in the population of patients undergoing orthotopic liver transplantation, many of whom have viral hepatitis.

References 1. LaBerge JM, Ring EJ, Gordon RL,

e t al. Creation of transjugular in- trahepatic portosystemic shunts with the Wallstent endoprosthesis: results in 100 patients. Radiology 1993; 187:413420.

2. Nakamura S, Tzuzuki T. Surgical anatomy of the hepatic veins and the inferior vena cava. Surg Gynecol Obstet 1981; 152:43-50.

3. Cohen GS, Ball DS. Delayed Wall- stent migration after a transjugular intrahepatic portosystemic shunt procedure: relocation with a loop snare. JVIR 1993; 4:561-563.

4. Cekirge S, Foster RG, Weiss JP , McLean GK. Percutaneous re-

moval of an embolized Wallstent during a transjugular intrahepatic portosystemic shunt procedure. JVIR 1993; 4:559-560.

5. LaBerge JM, Ferrell LD, Ring EJ, Gordon RL. Histopathology of TIPS stenoses and occlusions. JVIR 1993; 4:779-786.

6. LaBerge JM, Ferrell LD, Ring EJ, et al. Histopathologic study of transjugular intrahepatic portosys- temic shunts. JVIR 1991; 2:549- 556.

7. Palmaz JC. Intravascular stents: tissue-stent interactions and design considerations. AJR 1993; 160:613- 618.

8. Chervu A, Moore WS. An overview of intimal hyperplasia. Surg Gyne- col Obstet 1990; 71:433-447.

9. Palmaz JC, Windeler S, Garcia F, Tio FO, Sibbitt RR, Reuter SR. Atherosclerotic rabbit aortas: ex- pandable intraluminal grafting. Ra- diology 1986; 160:723-726.

10. Woodle ES, Darcy M, White HM. Intrahepatic portosystemic vascular stents: a bridge to hepatic trans- plantation. ~Grgery l-993; 113:344- - .

351. 11. Schultz SR, Ring EJ. Limiting di-

ameter in transjugular intrahepatic portosystemic shunt (abstr). Radiol- ogy 1994; 193(P):130.