Transjugular Intrahepatic Portosystemic ShuntCreation in a Polycystic Liver Facilitated byHybrid Cross-sectional/Angiographic ImagingDaniel Y. Sze, MD, PhD, Norbert Strobel, PhD, Rebecca Fahrig, PhD, Teri Moore, Stephan Busque, MD, and

Joan K. Frisoli, MD, PhD

Polycystic liver disease (PCLD) has long been considered to represent a contraindication to transjugular intrahepaticportosystemic shunt (TIPS) creation, primarily because of the risk of hemorrhage. Three-dimensional (3D) navigationwithin the enlarged and potentially disorienting parenchyma can now be performed during the procedure with thedevelopment of C-arm cone-beam computed tomography, which relies on the same equipment already used forangiography. Such a hybrid 3D reconstruction-enabled angiography system was used for safe image guidance of aTIPS procedure in a patient with PCLD. This technology has the potential to expedite any image-guided procedurethat requires 3D navigation.

J Vasc Interv Radiol 2006; 17:711–715

Abbreviations: IVC � inferior vena cava, PCLD � polycystic liver disease, 3D � three-dimensional, TIPS � transjugular intrahepatic portosystemic shunt,2D � two-dimensional

THE adoption of transjugular intrahe-patic portosystemic shunt (TIPS) cre-ation for treatment of complications ofportal hypertension has been wide-spread. However, the patient popula-tion is heterogeneous, and anatomicand physiologic contraindications stillremain (1). Although successful TIPScreation has been reported in three pa-tients with hepatic fibrosis secondary

to polycystic liver disease (PCLD) (2–4), the potential for technical failure,cyst rupture, and hemorrhage stillcauses PCLD to be considered at leasta relative, if not absolute, contraindi-cation to TIPS creation (1).

Several reports have described hy-brid projectional and cross-sectionalimaging methods to guide TIPS proce-dures, including the use of magneticresonance (MR) imaging (5), three-di-mensional (3D) ultrasound (US) (6),and intravascular US (7). All thesemethods require multiple distinct im-aging instruments. The developmentof digital flat-panel x-ray detectorswith high readout rates facilitates 3Dtomographic reconstruction, resultingin a new class of 3D-enabled hybridC-arm systems capable of produc-ing conventional projectional angio-graphic images as well as computedtomography (CT)–like images withuse of the same equipment. We usedsuch a hybrid instrument to guide aTIPS procedure in a patient with a se-verely enlarged and distorted liver af-fected by PCLD and fibrosis.

CASE REPORT

Our institutional review board doesnot require approval for this type ofretrospective case report.

A 62-year-old woman with an 18-year history of symptomatic polycys-tic kidney disease, PCLD, and hyper-tension presented with progressiveabdominal distension, pain, and re-fractory hypertension. She was func-tionally anuric and had an operationalBrescia-Cimino shunt. The patient un-derwent elective bilateral radical ne-phrectomies for relief of bulk symp-toms and renovascular contribution tohypertension. Although her immedi-ate postoperative course was uncom-plicated, she experienced severe bilat-eral lower extremity and lower truncalpitting edema as well as massive as-cites within 3 months.

MR imaging revealed near-occlu-sion of the inferior vena cava (IVC)and marked distortion of the hepaticveins caused by deforming cysts (Fig1). Hepatic venography and IVCvenography were performed, confirm-ing these findings and revealing gra-

From the Departments of Radiology (D.Y.S., N.S.,R.F., J.K.F.) and Transplant Surgery (S.B.), StanfordUniversity Medical Center, H-3646, Stanford, Cali-fornia 94305-5642; and Siemens Medical Solutions(N.S., T.M.), Hoffman Estates, Illinois. Received Oc-tober 5, 2005; revision requested; final revision re-ceived and accepted January 9, 2006. Address cor-respondence to D.Y.S.; E-mail: dansze@stanford.edu

The instrumentation portion of this project was sup-ported by NIH R01 EB003524, Siemens Medical So-lutions, and the Lucas Foundation. N.S. and T.M. areemployees of Siemens Medical Solutions. None ofthe other authors have identified a conflict of in-terest.

© SIR, 2006

DOI: 10.1097/01.RVI.0000208984.17697.58

711

dients of 8 mm Hg between the he-patic vein and the right atrium and 7mm Hg between the infrahepatic IVCand the right atrium. Self-expandingnitinol stents were deployed in theIVC stenosis and the middle hepaticvein, which was the only hepatic veinthat could be recognized and cannu-lated. Within approximately 3 weeks,during which there were nine hemo-dialysis sessions, the peripheraledema completely resolved. However,massive ascites persisted, requiringtherapeutic large-volume paracentesisof 8–9 L every 3 weeks. The asciteswas presumed to be a result of pro-gressive hepatic fibrosis with a possi-ble component of small-vessel hepaticvenous obstruction, but histopatho-logic studies were unavailable becausethe risk of hemorrhage from biopsywas considered prohibitive. On the ba-sis of a consensus among a transplantsurgeon, a nephrologist, an interven-tional radiologist, the patient, and thepatient’s family, we performed portaldecompression via creation of a TIPS.The patient’s Model for End-stageLiver Disease score was 20 as a resultof renal failure, and the Child-Pugh-Turcotte score was 8.

The patient was administered gen-eral anesthesia for maximum controlof respiration and motion. Projectional

images were acquired with use of aflat-panel C-arm angiography instru-ment (Axiom Artis dBA; Siemens

Medical Solutions, Forchheim, Ger-many). Cross-sectional images werereconstructed on a workstation (X-

Figure 1. Transverse axial T1-weighted MR images of the liver. (a) At the level of the confluence of hepatic veins, the IVC appears asa slit (arrow), and the only recognizable hepatic vein is a severely distorted middle vein (white arrowhead). The azygos system (blackarrowhead) is hypertrophied, likely as a result of IVC obstruction. (b) The main portal vein is patent (arrow) but surrounded by cystsof different signal intensities, representing fluids with different concentrations of protein and blood. Intrahepatic collateral hepaticvenous vessels (arrowheads) likely result from long-standing venous outflow obstruction.

Figure 2. A C-arm axial transverse CT image through the target main right portal vein(arrowheads) shows pericystic and aortic calcifications used as landmarks. This cross-sectional image was reconstructed with use of DynaCT software. The associated two-dimensional (2D) x-ray input projections were acquired with an Axiom Artis dBAflat-panel detector C-arm system.

712 • TIPS Creation in a Polycystic Liver with Hybrid Imaging April 2006 JVIR

Leonardo; Siemens) with use of 3D re-construction software (DynaCT; Sie-mens). No intravenous contrast agentwas administered. The cross-sectionalimages depicted the long and cyst-laden path between the hepatic veinand portal vein (Fig 2). Anatomiclandmarks, including individual cystsand calcifications, were identifiedfrom CT and MR images previouslyobtained with use of contrast mediumenhancement, and the relative posi-tion of the target right portal vein wasestablished in relation to these land-marks.

Access was obtained into the preex-isting middle hepatic vein stent with aHawkins TIPS set (Angiodynamics,Queensbury, NY). The distal 2 cm ofthe 22-gauge needle was custom-shaped with a gentle 15° curve to al-low fine steering. An initial pass wasperformed under fluoroscopic guid-ance, and 3D image reconstructionwas repeated (Fig 3). On the basis ofthe 3D data obtained, the needle wasthen advanced more inferiorly, anteri-orly, and medially, directly into theright main portal vein. The tract was

dilated to 8 mm in diameter and linedwith overlapping covered stents 10 cmand 9 cm long and 8 mm in diameter(Viatorr; W.L. Gore & Associates, Flag-staff, AZ). Digital subtraction angiog-raphy and 3D image reconstructionwith and without contrast agent injec-tion were performed, confirming suc-cessful creation of the shunt (Figs 4, 5)and revealing a small amount of intra-cystic hemorrhage without active ex-travasation. The portosystemic gradi-ent measured between the splenic veinand right atrium was reduced from 13mm Hg to 4 mm Hg.

Postoperatively, the patient showedstable vital signs and tolerated routinehemodialysis. Her hematocrit level in-creased 2% compared with measure-ments before the TIPS procedure. Shewas treated for spontaneous bacterialperitonitis in the first postoperativeweek, required one therapeutic para-centesis of 3 L in the first postopera-tive month, and requires low-dose lac-tulose administration for preventionof encephalopathy, but is now free ofascites at 6-month follow-up.

DISCUSSION

TIPS creation has essentially com-pletely replaced surgical portosys-temic shunt creation for treatment ofcomplications of portal hypertension.The original consensus documentpublished by the National DigestiveDiseases Advisory Board in 1995 (8)outlined indications and contraindica-tions, but many guidelines were basedon skepticism and theoretical riskswithout scientific evidence. The up-dated guidelines from the AmericanAssociation for the Study of Liver Dis-eases (1) show that some of the clinicalcontraindications continue to be ac-cepted (eg, heart failure, severe pul-monary hypertension), but technicaland anatomic contraindications suchas Caroli disease, portal vein thrombo-sis, and PCLD need to be reconsid-ered, especially since the advent ofcovered stents. TIPS creation in thepatient with polycystic kidney diseasewithout hepatic cysts is technicallyroutine (9), but the safe decompressionof portal hypertension in patients withhepatic cysts depends largely on the

Figure 3. Repeat cross-sectional reconstructed images after the first advancement of TIPS needle. (a) Axial transverse image at the samelevel as Figure 2 shows the tip of the needle lying approximately 2 cm posterolateral to the target right portal vein (arrow). (b) Coronalreconstruction along the needle path shows the guiding cannula within the preexisting middle hepatic vein stent and the 22-gaugeneedle traversing several large cysts toward the landmark calcifications. The needle was adjusted from this position to puncture thetarget.

Sze et al • 713Volume 17 Number 4

ability to locate the portal vein amidthe array of cysts (2–4).

The development of ascites afternephrectomy in patients with polycys-tic kidney disease has been reported inthree patients and is thought to becaused by hepatic venous outflow ob-struction from torsion and distortionof the hepatic veins and IVC from therepositioning of the liver (10,11). Pa-tients may also experience hepatic fi-brosis with or without cystic involve-ment of the liver, which can eventuallylead to ascites. Stent implantation ofthe IVC in our patient relieved lower-extremity edema, but stent implanta-tion of the dominant hepatic vein wasineffective in the treatment of the as-cites, suggesting that macroscopic ob-struction of venous outflow was notthe sole cause of ascites. The relativelylow portosystemic gradient measuredon portal vein puncture (13 mm Hg)and the minimal varices observed raisedfurther questions about the actual causeof the ascites, but we proceeded withthe shunt procedure because predicting

efficacy on the basis of the gradient be-fore TIPS creation is difficult (12), andTIPS creation is potentially reversible.The patient’s positive clinical responseto TIPS creation suggests that portal hy-pertension played a role.

Three-dimensional image recon-struction with use of a C-arm systemgenerates an isotropic 3D data setfrom 2D x-ray input projections. Fordata acquisition, we used the 30-cm �40-cm detector of the Axiom ArtisdBA biplane system. Acquisition pro-tocols that require a maximum of 20seconds result in as many as 538 pro-jections through a partial rotation of atleast 200°. For optimal detector perfor-mance, this system relies on a dose-control system whereby the detectorentrance dose is held constant by ad-justment of the current time product ofthe x-ray tube and regulation of thetube voltage only if needed. In ourcase, 20-second runs were performedwith the x-ray source voltage at 125kVp and the detector entrance dose setat 1.2 �Gy per projection.

The 2D x-ray projections are subse-quently fed into a 3D tomographic im-age reconstruction algorithm modifiedto account for irregular but stable scantrajectories (13). To optimize 3D soft-tissue image quality, a sequence ofcorrection algorithms is applied, in-cluding scatter correction, beam-hard-ening correction, truncation correc-tion, and ring-artifact correction (14).The enhanced 3D reconstruction algo-rithm, marketed as DynaCT, also pro-vides bone and vessel kernels for 3Dreconstruction. Bone kernels are de-signed for tomographic reconstructionof soft-tissue details, whereas vesselkernels are optimized for contrastmedium–enhanced vasculature. Soft-tissue structures such as cysts inside aliver are best depicted by bone kernelsto avoid the inherent edge enhancementof vessel kernels. We reconstructed 3Ddata sets with an isotropic voxel size ofapproximately 0.8 mm. Voxels in neigh-boring slices may be averaged toachieve a larger effective slice width andthus a reduction in noise.

Figure 4. Digital subtraction angiography performed on the same C-arm device used to acquire the 2D images used for 3D C-arm CTreconstruction. (a) Initial portal venogram shows sluggish antegrade flow into the distorted intrahepatic portal system and someretrograde flow into mesenteric vessels. Note that the length of the tract from the portal vein wall to the IVC measures more than 13cm, and the needle throw from the hepatic vein wall to the portal vein wall measures more than 9 cm. (b) Portal venogram afterplacement of overlapping stents 10 cm and 9 cm long and 8 mm in diameter shows some preservation of intrahepatic portal flow andno extravasation. The portosystemic gradient is reduced from 13 mm Hg to 4 mm Hg.

714 • TIPS Creation in a Polycystic Liver with Hybrid Imaging April 2006 JVIR

The use of a hybrid 2D/3D imaginginstrument capable of 2D angio-graphic projection imaging and 3D re-construction facilitated our long trans-cystic TIPS creation procedure andminimized blood loss. Considerableextra planning was required beforethis challenging procedure was at-tempted, and successful use of thisnew technology will likely remainvery operator dependent. With use ofaccurate imaging and impermeablecovered stents, TIPS creation in thepatient with PCLD appears to be safeand effective.

Acknowledgments: The authors thankMervin Nicolas for outstanding technicalsupport.

References1. Boyer TD, Haskal ZJ. American Asso-

ciation for the Study of Liver Diseasespractice guidelines: the role of trans-jugular intrahepatic portosystemicshunt creation in the management ofportal hypertension. J Vasc Interv Ra-diol 2005; 16:615–629.

2. Shin ES, Darcy MD. Transjugular in-trahepatic portosystemic shunt place-ment in the setting of polycystic liverdisease: questioning the contraindica-tion. J Vasc Interv Radiol 2001; 12:1099–1102.

3. Bahramipour PF, Festa S, Biswal R, etal. Transjugular intrahepatic porto-systemic shunt for the treatment of in-tractable ascites in a patient with poly-cystic liver disease. Cardiovasc IntervRadiol 2000; 23:232–234.

4. Spillane RM, Kaufman JA, Powelson J,et al. Successful transjugular intrahe-patic portosystemic shunt creation in apatient with polycystic liver disease.AJR 1997; 169:1542–1544.

5. Kee ST, Ganguly A, Daniel BL, et al.MR-guided transjugular intrahepaticportosystemic shunt creation with useof a hybrid radiography/MR system. JVasc Interv Radiol 2005; 16:227–234.

6. Rose SC, Behling C, Roberts AC, et al.Main portal vein access in transjugularintrahepatic portosystemic shunt pro-cedures: use of three-dimensional ul-trasound to ensure safety. J Vasc IntervRadiol 2002; 13:267–273.

7. Petersen B. Intravascular ultrasound-guided direct intrahepatic portacavalshunt: description of technique andtechnical refinements. J Vasc Interv Ra-diol 2003; 14:21–32.

8. Shiffman ML, Jeffers L, Hoofnagle JH,et al. The role of transjugular intrahe-patic portosystemic shunt for treat-ment of portal hypertension and itscomplications: a conference sponsoredby the National Digestive Diseases Ad-visory Board. Hepatology 1995; 22:1591–1597.

9. Johnson SP, Leyendecker JR, JosephFB, et al. Transjugular portosystemicshunts in pediatric patients awaitingliver transplantation. Transplantation1996; 62:1178–1181.

10. Dionisio P, Sessa A, Conte F, et al.Budd-Chiari syndrome following pre-transplant mononephrectomy in an au-tosomal dominant polycystic kidneydisease patient with liver cysts.Nephron 1997; 75:109–111.

11. Clive DM, Davidoff A, Schweizer RT.Budd-Chiari syndrome in autosomaldominant polycystic kidney disease: acomplication of nephrectomy in pa-tients with liver cysts. Am J Kidney Dis1993; 21:202–205.

12. Nair S, Singh R, Yoselewitz M. Cor-relation between portal/hepatic veingradient and response to transjugularintrahepatic portosystemic shunt cre-ation in refractory ascites. J Vasc IntervRadiol 2004; 15:1431–1434.

13. Wiesent K, Barth K, Durlak P, et al.Enhanced 3-D reconstruction algo-rithm for C-arm systems suitable forinterventional procedures. IEEE TransMed Imaging 2000; 19:391–403.

14. Zellerhoff M, Scholz B, RuehrnschopfEP, et al. Low contrast 3D reconstruc-tion from C-arm data. SPIE Med Imag-ing Symp 2005;5745–5771.

Figure 5. Repeat transverse axial C-arm CT image obtained after completion of the TIPSprocedure several centimeters more cephalad than in Figures 2 and 3a. A small amountof intracystic hemorrhage is observed (arrowheads) involving a traversed cyst, which prob-ably occurred during tract dilation. Delayed images (not shown) reveal no change, indicatingno active extravasation. The stent anterior to the vertebral body is within the IVC.

Sze et al • 715Volume 17 Number 4