Longtermfollow transjugularintrahepatic stent shunt (TIPSS ...TABLEIil Results ofrebleeding, encephalopathy, ascites, andsepsis after TIPSS Variceal rebleeding 24 episodesin 16 patients

Gut 1996; 39: 479-485

Longterm follow up of transjugular intrahepaticportosystemic stent shunt (TIPSS) for thetreatment of portal hypertension: results in 130patients

A J Stanley, R Jalan, E H Forrest, D N Redhead, P C Hayes

Departments ofMedicineA J StanleyR JalanE H ForrestP C Hayes

and RadiologyD N Redhead

Royal Infirmary ofEdinburgh, EdinburghCorrespondence to:Dr AJ Stanley,Department of Medicine,Royal Infirmary ofEdinburgh, Lauriston Place,Edinburgh, EH3 9YW.Accepted for publication27 February 1996

AbstractBackground-Transjugular intrahepaticportosystemic stent shunts (TIPSS) areincreasingly being used to manage thecomplications of portal hypertension.This study reports on the follow up on 130patients who have undergone TIPSS.Patients and Methods-One hundred andthirty patients (81 male), mean (SD) age54.7 (12.5) years underwent TIPSS. Themajority (64.6%) had alcoholic cirrhosisand 53.2% had Childs C disease. Indi-cations were: variceal haemorrhage(76.2%), refractory ascites (13.1%), portalhypertensive gastropathy (4.6%), others(6.1%). Shunt function was assessed byDoppler ultrasonography and two then sixmonthly portography and mean follow upfor survivors was 18-0 months (range2-43.5).Results-The procedure was successful in119 (91.5%). Sixty three episodes of shuntdysfunction were observed in 45 (37.8%)patients. Variceal rebleeding occurred in16 (13.40/o) patients and was always asso-ciated with shunt dysfunction. Twenty(16.8%) patients had new or worse spon-taneous encephalopathy after TIPSS and11 (64.7%) patients had an improvementin resistant ascites. Thirty day mortalitywas 21.8% and one year survival 62.5%.Conclusion-TIPSS is an effective treat-ment for variceal bleeding, resistantascites, and portal hypertensive gastro-pathy. Rebleeding is invariably associatedwith shunt dysfunction, the frequency ofwhich increases with time, thereforeregular and longterm shunt surveillance isrequired.(Gut 1996; 39: 479-485)

Keywords: transjugular intrahepatic portosystemicstent shunt, portal hypertension, varices, ascites.

Since their introduction into clinical practice in1989,' transjugular intrahepatic portosystemicstent shunts (TIPSS) are being increasinglyused in the management of both varicealhaemorrhage and refractory ascites.

Variceal haemorrhage is the most dramaticcomplication of portal hypertension, occurringin 30% patients with cirrhosis during theirlifetime.2 Mortality from the first bleedapproaches 50%3 and 70-100% patients have

recurrent bleeding. Immediate control ofbleeding can be achieved in 90% of patients byballoon tamponade,4 vasoactive drug therapy,5sclerotherapy,6 variceal band ligation7 orsurgery,8 9 with the greatest reduction in re-bleeding rates achieved by surgical shunts.'0The main limitations of shunt surgery are itshigh perioperative mortality and frequency ofdebilitating postoperative encephalopathy,which approaches 30% in some series." TIPSShas been shown to control active varicealhaemorrhage and reduce rebleeding rates whilehaving lower procedure related complicationsand probably less post-treatment encephalo-pathy'2 compared with surgical shunts. Followup of patients with TIPSS is however limited,which is important as encephalopathy follow-ing surgical shunts commonly was delayed andshunt patency after TIPSS may decrease withtime.

Refractory ascites is associated with ad-vanced liver disease and a poor prognosis.'3The current therapeutic options of repeatedparacentesis and a peritoneovenous shunt arefar from ideal and associated with significantmorbidity and prolonged hospital stay and donot affect survival.'4 TIPSS permits bettercontrol of ascites and improves renal sodiumexcretion'5 although the mortality seemsunchanged in the limited reported data. TIPSShas also been used and found to be effective inthe control of other conditions such as portalhypertensive gastropathy, splenomegaly,'6Budd-Chiari syndrome,'7 and hepatic hydro-thorax.'8 The present available literature islimited both by short duration of follow up andsmall patient numbers. Up until August 1995,we had carried out 130 TIPSS procedures inour unit and the aim of this paper is to presentour results and more importantly the longtermfollow up.

Methods

PATIENTSFrom 1991 to 1995, TIPSS insertion wasattempted in 130 patients, with successfulplacement of the stent in 119 cases (91.5%).Table I shows the details of the patients. Eightyone patients were male with mean (SD) age of54.7 (12.5) years (range 9 to 83 years). Overthe same period, a total of 220 patients weretreated for variceal haemorrhage at ourinstitution.

479

on May 22, 2021 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.39.3.479 on 1 S

eptember 1996. D

ownloaded from

http://gut.bmj.com/

Stanley, _Jalan, Forrest, Redhead, Hayes

TABLE i Patient characteristics (n=130) before TIPSS

Sex (M/F) 81/49

Age (y)mean (SD) 54-7 (12.5)range 9-83

Aetiology of liver disease (/)alcoholic cirrhosis 84 (64 6)cryptogenic cirrhosis 12 (9.2)primary biliary cirrhosis 11 (8.5)hepatitis C 5 (3.8)hepatitis B 5 (3.8)cystic fibrosis 3 (2.3)primary sclerosing cholangitis 2 (1.5)autoimmune hepatitis 2 (1.5)other 6 (4.6)

Childs-Pugh score (mean (SD)) 9.9 (2.7)Childs-Pugh grade (/)A 10 (7.9)B 49 (38.9)C 67 (53.2)

Indication (%)oesophageal varices 73 (56.2)gastric varices 26 (20.0)refractory ascites 17 (13.1)portal hypertensive gastropathy 6 (4.6)ectopic vances 5 (3.8)other 3 (2.3)

Emergency procedure (%) 35 (26.9)Clinical features (%)

artificially ventilated 19 (14.6)haemodynamic compromise 10 (7.7)ascites 86 (66.2)hepatic encephalopathy 44 (33 8)

The aetiology of portal hypertension wasalcoholic cirrhosis (ALD) in 84 (64.6%)patients, with other aetiologies shown inTable I. One patient had al antitrypsindeficiency and one Budd-Chiari syndrome.Four patients were non-cirrhotic (one eachwith amyloid, idiopathic portal hypertension,polycystic disease, and nodular regenerativehyperplasia). Most cirrhotic patients (53.2%)had Childs-Pugh grade C disease at time ofTIPSS with a mean (SD) overall Childs-Pughscore of 9 9 (2.7).The indication for TIPSS was oesophageal

variceal bleeding in 73 (56-2%) patients whohad either continued bleeding despite twosessions of sclerotherapy or were part of a trialcomparing band ligation with TIPSS in theprevention of rebleeding (see Table I). Theindication was ectopic varices in five (3.8%)patients (two rectal, two stomal, and one duo-denal) and painful splenomegaly, hypersplen-ism and embolisation of a spontaneous shunt inone patient each. Thirty five procedures(26.9%) were carried out as an emergency with19 (14-6%) patients receiving assisted ventila-tion and 23 (17.7%) patients treated withballoon tamponade prior to TIPSS.

STUDY DESIGN

The technique of TIPSS placement was basedon the original method described by Richter`9and is described in detail elsewhere.20 Routinepre-procedural mesenteric angiography wasundertaken in the first 32 patients to guideportal vein puncture and in 27 subsequentpatients, Doppler ultrasonography was used toidentify the site of portal vein bifurcation. Inthe last 71 patients however, no routineimaging was undertaken to localise the portalvein pre-TIPSS, although ultrasonography wasused prior to the procedure to exclude portalvein thrombosis. These changes have evolvedas a result of ongoing audit at our unit.

Once successful puncture of the portal veinwas achieved, 2-3 Palmaz stents (Johnson andJohnson) (20 patients), or 1-2 Wallstents(Schneider US Stent Division) (99 patients)were inserted to reduce the portal pressuregradient (defined as: (portal pressure) -(inferior vena-caval pressure)) to less than 12mm Hg. Three patients subsequently hadAngiomed stents (Angiomed, Karlsruhe,Germany) inserted to reduce the shunt size. Intwo patients in whom a thrombus was notedwithin the portal vein at the end of theprocedure, a catheter was left within the shuntfor regional infusion of low dose streptokinasefor 24 hours. Prophylactic antibiotics (cefo-taxime and amoxycillin) were given one hourbefore the procedure and continued for 48hours thereafter and Doppler ultrasonographywas performed prior to discharge to ensureshunt patency. Routine portography wasundertaken at one to three months and sixmonthly thereafter to assess shunt function, orearlier in the event of rebleeding or reaccumu-lation of ascites. Early in our experience how-ever, several shunts were left six to 12 monthsbefore initial angiographic assessment.

Encephalopathy was assessed clinically aftersix weeks then at three monthly intervalsduring outpatient review. Prophylactic lactu-lose and protein restriction were not routinelyapplied. Variceal rebleeding was defined asendoscopically confirmed variceal haemor-rhage occurring more than 24 hours afterTIPSS insertion. All shunt complications wereconfirmed angiographically with occlusiondefined as absent flow through the shunt.Pseudo-intimal hyperplasia, hepatic veinstenosis, portal vein and shunt thrombosiswere defined by the angiographic appearancein conjunction with either a 20% rise in portalpressure gradient or an increase in portalpressure gradient to 12 mm Hg or more.Primary shunt patency was defined as the (pre-intervention) absence of any ofthe above shuntcomplications. Mean (SD) follow up (definedas time to death, most recent clinical review orliver transplantation) for all patients was 107(11.0) months and for survivors (patients aliveup to October 1995) 18-0 (1 19) months.

STATISTICAL ANALYSISResults are expressed as mean (SD) or rangewhere indicated. Paired Student's t test wasused to determine statistical significance andKaplan-Meier method used for rates ofvaricealrebleeding, primary shunt patency, and survival.

ResultsTables II and III summarise the results.

SHUNT PROCEDURETIPSS placement was successful in 119(91.5%) patients. The procedure failed in 11patients (10 of whom had variceal haemor-rhage) because a main branch ofthe portal veincould not be punctured: four of these patientssubsequently underwent shunt surgery and five

480

on May 22, 2021 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.39.3.479 on 1 S

eptember 1996. D

ownloaded from

http://gut.bmj.com/

TIPSSfor the treatment ofportal hypertension

TABLE Ii Results of TIPSS procedure and shuntcomplications

Successful TIPSS procedure (/) 119 (91.5)PPG before TIPSS mean (SD) (mm Hg) 17-5 (5.9)PPG after TIPSS (mm Hg) 8-6 (3.4)Pre-intervention shunt patency ():

6 months 47 (71-2)1 year 25 (58-1)2 years 3 (214)

Shunt complications: 63 episodes in45 (37 8) patients:

pseudo-intimal hyperplasia 29hepatic vein stenosis 13portal vein/shunt thrombosis 5occlusion 1 6

PPG=portal pressure gradient.

had further endoscopic therapy. Thirty daymortality in the failed TIPSS group withvariceal haemorrhage was 70.0%.Two procedure related deaths occurred from

intraperitoneal haemorrhage because of anextrahepatic tear of the portal vein in one andpuncture of the liver capsule in the other. Onepatient developed an epidural haemorrhage,which was diagnosed after the procedure butthe exact relation to TIPSS placement wasunclear. Other complications included portalvein thrombosis in two patients (successfullytreated by local streptokinase infusion), portalvein dissection in one (successfully managed bystenting), and shunt dislocation or migrationinto the splenic vein in two. There were noclinically significant groin or neck haema-tomas. Mean (SD) portal pressure gradient

TABLE Iil Results of rebleeding, encephalopathy, ascites, and sepsis after TIPSS

Variceal rebleeding 24 episodes in 16 patients (13-4%)Non-variceal rebleeding 9 patients (7.6%)

7 sclerotherapy ulcers1 duodenal ulcer1 Mallory-Weiss tear

Encephalopathy (%)before TIPSS 44 (33.8)new/worse spontaneous encephalopathy 20 (16-8)new/worse encephalopathy, secondary to sepsis 9 (7.6)new/worse encephalopathy, secondary to bleeding 6 (5.0)

Ascites (%)before TIPSS 86 (66.2)improved after TIPSS 56 (65.1)reaccumulation 14 (25.0)primary indication for TIPSS 17 (13-1)

Sepsis after TIPSS (%) 14 (11-8)

100

80

60

40

20

o 10 20 30Follow up after TIPSS (months)

Figure 1: Kaplan-Meier analysis ofpatients free of variceal rebleeding and with primaryshunt patency duringfollow up.

was reduced from 17.5 (5.9) mm Hg pre-TIPSS to 8.6 (3.4) mm Hg (p

Stanley, Jalan, Forrest, Redhead, Hayes

AscitesEighty six patients (66&2%) had ascites prior toTIPSS and this was the primary indication in17 patients (seven of whom had establishedhepatorenal syndrome) (see Table III). Theascites improved (reduced or no diureticrequirement) in 56 (65. 1O%) of the 86 patients,but reaccumulated in 14. This was associatedwith shunt insufficiency in 11 patients andspontaneous bacterial peritonitis in two, butwas responsive in all cases to shunt revision andantibiotics respectively. Of the 17 patients whohad TIPSS performed for refractory ascites,improvement occurred in 11 and two havesubsequently undergone successful liver trans-plantation. Those with biochemical evidenceof renal dysfunction tended to respond less wellafter TIPSS insertion.

Other indicationsTIPSS was performed for intractable bleedingfrom portal hypertensive gastropathy in sixpatients. Four of these have not requiredfurther transfusion and the other two haverequired one admission each for transfusion.Both of these patients had evidence of shuntdysfunction due to pseudo-intimal hyperplasiain association with a raised portal pressuregradient, which was successfully treated byangioplasty or shunt extension. TIPSS resultedin an improvement in platelet count from13 000/1 to 50 000/1 in one patient with hyper-splenism. Another patient had a largespontaneous shunt successfully embolised viathe TIPSS for amelioration of intractablehepatic encephalopathy.

Hepatic encephalopathyForty four (33.8%) patients were clinicallyencephalopathic prior to TIPSS and thisresolved in 24 (54.5%) after the procedure.Twenty (16.8%) patients developed new orworsening spontaneous encephalopathy duringfollow up (15 within the first six months). Afurther 15 (12.6%) patients developed en-cephalopathy secondary to sepsis or bleedingduring follow up (see Table III). Reduction inshunt size was performed because of en-cephalopathy in four patients (successful inthree), with all others responding to simplemedical treatment.

Childs gradeC-B-A--

1.0i7 0.91

0.80.7 \

" 0.6> 0.5cc 0.4= 0.3E 0.2U 0.1

0 10 20 30 40 50Follow up after TIPSS (months)

Figure 2: Kaplan-Meier analysis ofpatient survival afterTIPSS by Childs grade. (Childs A, n=12; Childs B, n=44;Childs C, n=61. At 12 months, n=9, 23, and 14respectively.)

SepsisFourteen (11 .8%) patients developed clinicallysignificant infections in the week after TIPSS(seven pulmonary, two spontaneous bacterialperitonitis, two related to central venouscatheter, one cellulitis, and two of unknownorigin). Sepsis was the cause of death in twopatients during the index admission: one withpre-existing staphylococcal and fungal septi-caemia and one with cystic fibrosis and pre-existing lung sepsis. All other infective episodesresponded to antibiotics.

LiverfunctionApproximately one third of patients exhibiteda transient deterioration in their liver functiontests in the first week after TIPSS, usually withdoubling of bilirubin and alanine amino-transferase values. Three patients died withclinical features of acute liver failure, character-ised by hypotension, renal failure, and hypo-glycaemia. Two of these patients also hadevidence of raised intracranial pressure onextradural pressure recordings or computedtomography. Childs-Pugh score did howeverimprove at three to six months to 7-6 (2. 1)(p=0.019) largely because of a reduction inascites.

TransplantationThirteen patients have undergone orthotopicliver transplantation after TIPSS. Transplantwas undertaken a mean of 7.1 (5.4) monthsafter TIPSS and 11 of these patients remainalive and well.

MortalityOf the 119 patients with successful TIPSSplacement, 52 have died and 13 have under-gone transplantation. Procedure relatedmortality was 1.5% and mean time to deathwas 6-0 (8.9) months (range 0.03-45.3). Meanfollow up of survivors is 18.0 (11.9) months(range 2 0-43.5). Thirty day mortality was21.8% (84/6% ofwhom had Childs C disease)and six months survival 69.2%. One and twoyear survival is 62.3% and 46.5% respectively.Mortality was dependent on Childs grade attime of TIPSS (see Fig 2).

DiscussionThis is only the second study with largenumbers of patients and longterm follow upand our results are similar to the two yearfollow up of 90 patients reported by Labergeet al.2' Shunt insertion was achieved with areduction in the portal pressure gradient tobelow 12 mm Hg in 91.5% patients. Ourrecurrent variceal rebleeding rate (13.4%patients) and rate of new or worseningspontaneous encephalopathy (16.8% patients)compares favourably with most otherseries.12 21-23TIPSS has established a position as a 'rescue

procedure' for uncontrolled variceal bleeding.Of the 35 patients who had TIPSS for

482

on May 22, 2021 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.39.3.479 on 1 S

eptember 1996. D

ownloaded from

http://gut.bmj.com/

TIPSSfor the treatment ofportal hypertension

continued variceal haemorrhage despite sclero-therapy with or without balloon tamponade,shunt creation was successful in the vastmajority (914%). Most (77.1%) of thesepatients had Childs C cirrhosis. Although onlytwo patients had continued bleeding theinpatient mortality in this high risk group was43.7%. The options of a surgical portosystemicshunt or transection in this situation havehigher procedure related and early mortality.24In addition transection has higher rebleedingrates2' and surgical shunting has higher rates ofencephalopathy26 while also compromising asubsequent liver transplant. It should be notedthat in many cases TIPSS is performed incritically ill patients who would not be con-sidered fit for surgery.

Similar to other investigators,21 we foundshunt dysfunction in all patients with varicealrebleeding. There seems to be a steady attritionrate in shunt patency with time (Table II andFig 1) with a primary (pre-intervention)patency of 21.4% at two years. We have pre-viously shown that shunt dysfunction is morecommon with Palmaz stents than Wallstents,27although it should be noted that our early (oftenPalmaz stented) patients were left longer thanthree months prior to initial angiographicassessment, therefore especially in this group itis difficult to be sure of the exact timing of shuntdysfunction. Accepting this limitation, if it isbelieved important to assess the TIPSS prior toa 20% risk of dysfunction, it is necessary tocheck the shunt at two months and thenapproximately six monthly thereafter. It wouldseem that early shunt dysfunction is more likelyto be associated with variceal rebleeding thanthat occurring later after TIPSS, therefore evencloser shunt assessment in the first few monthsafter placement may be neccessary.Although intervention maintained shunt

patency in all but one patient when insuf-ficiency was detected, most stents treated withballoon dilatation required repeat dilatation onfollow up. This procedure would seem to haveonly a temporary effect and the longtermbenefits of and exact indications for dilatationremain unclear. The current limitations ofultrasonographic methods of shunt assess-ment28 mean that direct portography remainsthe 'gold standard' for surveillance. This is animportant limitation of TIPSS but hopefullythe development of new covered stents willtackle this problem.An ability to identify early shunt dysfunction

and minimise encephalopathy together with alow incidence of rebleeding makes TIPSS analternative to sclerotherapy in the treatment ofrecurrent variceal bleeding. Trials are currentlyunderway to compare these two treatments inthis situation. In the case of gastric varicealhaemorrhage where current endoscopic treat-ments are unsatisfactory, we believe TIPSS tobe the treatment of choice, as indicated by thecomparatively large proportion of patients inour series with gastric varices.Nine (7.6%) patients developed non-

variceal bleeding during follow up, mostly as aresult of sclerotherapy ulcers and these werethe cause of death in two patients. Sclero-

therapy ulcers have been recognised as a majorproblem by many investigators and remain amajor potential cause of morbidity andmortality no matter how effective the shunt is.The alternative of band ligation should reducethe frequency of this complication.29New or worsening spontaneous encephalo-

pathy was seen in 16S8% patients and a further12.6% had new or worsening encephalopathysecondary to sepsis or bleeding after TIPSS.Considering however that 33.6% patients wereencephalopathic prior to TIPSS, it is difficultto determine the exact role of the shunt in thedevelopment of encephalopathy in this group.In all but four patients (who went on to shuntreduction), the encephalopathy responded tosimple medical treatment such as lactulose andprotein restriction. This contrasts with theoften debilitating encephalopathy seen inaround one third of patients after non-selectivesurgical shunts"1 and is probably because of thesmall stent diameter and intrahepatic positionthat encourages continued portal flow into theliver.30 A reduced incidence of encephalopathyhas been reported in narrow portocavalH-grafts3' and selective splenorenal shunts32however the associated risks of surgery andsubsequent problems with transplantation mustagain be considered. Recent reports indicatethat increasing age and a previous history ofencephalopathy are the main predictors of en-cephalopathy after TIPSS33 and perhaps thesepatients should have narrower stents placed.Treatment of patients with refractory ascites

remains a major clinical problem. Apart fromliver transplantation, the main therapeuticoptions are repeated large volume paracentesisand peritoneovenous shunting, both of whichare associated with significant morbidity andprolonged inpatient management and mayaggravate functional renal failure. Recentreports'4 15 describe the benefits of TIPSS forrefractory ascites although data are limited. Themechanism of action seems to be a combina-tion of reduced portal pressure gradient andincreased natriuresis, probably secondary toincreased effective circulating plasma volumeand neurohumeral factors. Survival is poor inpatients with refractory ascites and any trialassessing TIPSS in this situation would need toassess quality of life and cost-benefit factors.The place of TIPSS in the treatment of

hepatorenal syndrome also remains contro-versial. Several authors have reported improve-ment in functional renal failure'5 34 and apreliminary study35 showed a decrease afterTIPSS in the levels of endothelin 1, which isthought to have an important role in thesyndrome. However as we found in this study,results are generally disappointing in this groupof patients and they have a very poor prognosiswithout liver transplantation.

Similar to other investigators,36 we found atransient deterioration in liver function tests ina third of patients probably secondary toreduced hepatic perfusion. Three patients diedsoon after TIPSS with clinical features of acuteliver failure, two of whom had evidence ofraised intracranial pressure. This is unusual inpatients with chronic liver disease but has been

483

on May 22, 2021 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.39.3.479 on 1 S

eptember 1996. D

ownloaded from

http://gut.bmj.com/

484 Stanley, _Jalan, Forrest, Redhead, Hayes

reported.37 All three of these patients wereextremely ill at the time of TIPSS, however,with prolonged hypotension and requiringventilation. The role of TIPSS in their clinicaldeterioration is therefore difficult to define,although any reduction in liver perfusion insuch patients may be critical. Overall howeverthere was a significant improvement in Childs-Pugh score at three to six months largely dueto improvement in ascites.

Procedure related mortality was 1.5%,which is similar to most reported series andcompares favourably with the operative mor-tality associated with surgical portosystemicshunts of around 10%.38 We found a relativelyhigh 30 day mortality of 21 8% compared withsome groups,22 however mortality has beenconsistently shown to depend on Childs gradeat time of TIPSS and our higher percentage ofChilds B and C patients (38.9% and 53-2%respectively) and higher numbers ofemergencyprocedures (26.9%) would account for this.Our one and two year survival of 62-3% and46.5% respectively are similar to the recentlyreported series from San Francisco.2'Longterm survival after TIPSS will depend

on the underlying severity of liver disease andimproved survival due to reduced rebleedingwill be difficult to detect. However the benefitsof reduced morbidity, as a result of improvedcontrol of ascites and less variceal rebleedingshould not be underestimated. We havepreviously shown that besides severity of liverdisease, hyponatraemia and encephalopathyprior to TIPSS independently determine long-term survival.39 Thirteen of our patients sub-sequently underwent liver transplantation andthe benefits of TIPSS as a 'bridge to trans-plantation' have been previously reported.40 Inparticular the ability to avoid surgery in thetreatment of variceal bleeding before trans-plant is important.

In conclusion, TIPSS is a comparativelysimple procedure that can be successfully per-formed even on critically ill patients. It iseffective in the treatment of both acute andrecurrent variceal haemorrhage (especiallywhere they are not amenable to sclerotherapy),refractory ascites, and portal hypertensivegastropathy. Unlike surgical portosystemicshunts, it has the advantage of not compro-mising subsequent liver transplantation. Shuntdysfunction however is common and seems toincrease linearly with time and variceal re-bleeding and reaccumulation of ascites usuallyoccur in the presence of shunt stenosis orocclusion. Regular surveillance by portographyis therefore necessary and this represents animportant limitation ofTIPSS. Post-procedureencephalopathy is significant but generallyeasily treated and bleeding from sclerotherapyulcers remains an important cause ofmorbidityand mortality. Trials are currently underway tocompare TIPSS with sclerotherapy or bandingin the treatment of first variceal haemorrhageand with paracentesis for refractory ascites,which will help define its exact role in thesesituations.We thank Fiona Miller for technical assistance and datacollection.

1 Richter GM, Noeldge G, Palmaz JC. The transjugularintrahepatic portosystemic stent-shunt (TIPSS): experi-ence of results of a pilot study. Cardiovasc Intervent Radiol1990; 13: 200-7.

2 Cales P, Pascal JP, Histoire naturelle des varices oeso-phagiennes au cours de la cirrhose (de la naissance a larupture). Gastroenterol Clin Biol 1988; 12: 245-54.

3 Graham D, Smith J. The course of patients after varicealhaemorrhage. Gastroenterology 1981; 80: 800-9.

4 Panes J, Teres J, Bosch J. Efficacy of balloon tamponade intreatment of bleeding gastric and oesophageal varices. DigDis Sci 1988; 33: 454-9.

5 Burroughs AK, McCormick PA, Hughes MD, Sprengers D,D'Heygere F, McIntyre R. Randomised double-blindcontrolled trial of somatostatin for variceal bleeding.Emergency control and prevention of early varicealrebleeding. Gastroenterology 1990; 99: 1388-95.

6 Westaby D, Hayes PC, Gimson AES, Polson RJ, Williams R.Controlled trial of ejection sclerotherapy for activevariceal bleeding. Hepatology 1989; 9: 274-7.

7 Stiegman GV, Goff JS, Michaletz-Onody PA, Korula J,Lieberman D, Saeed ZA, et al. Endoscopic sclerotherapyas compared with band ligation for bleeding oesophagealvarices. NEnglJrMed 1992; 326: 1527-32.

8 Burroughs AK, Hamilton G, Philips A, Mezzanotte G,McIntyre N, Hobbs KEF. A comparison of sclerotherapywith staple transection of the oesophagus for theemergency control of bleeding from oesophageal varices.NEnglJfMed 1989; 321: 857-62.

9 Editorial. Emergency portocaval shunts. (Anonymous).Lancet 1991; 337: 952.

10 Cello J, Grendall J, Crass R, Weber T, Trunkey D.Endoscopic sclerotherapy versus portocaval shunt inpatients with severe cirrhosis and acute varicealhaemorrhage: long term follow up. N Engl J Med 1987;316: 11-5.

11 Franco D, Smadja C. Prevention of recurrent varicealbleeding: surgical procedures. In: Benhamou JP,Lebrec D, eds. Clinics in gastroenterology: portalhypertnsion. London: WB Saunders, 1985: 233-57.

12 LaBerge JM, Ring EJ, Gordon RL, Lake JR, Docherty MM,Somberg KA, et al. Creation of transjugular intrahepaticportosystemic shunts with Wallstent endoprosthesis:results in 100 patients. Radiology 1993; 187: 413-20.

13 Moore K, Wilkinson S, Williams R. Ascites and renaldysfunction in liver disease. In: Millward-Sadler GH,Wright R, Arthur NJP, eds. Wright's liver and biliarydisease. London: Saunders, 1992: 1346-71.

14 Quiroga J, Sangro B, Nunez M, et al. Transjugular intra-hepatic portal-systemic shunt in the treatment ofrefractory ascites: Effect on clinical, renal, humeral andhaemodynamic parameters. Hepatology 1995; 21:986-94.

15 Ochs A, Rossle M, Haag MD, et al. The transjugularintrahepatic portosystemic stent-shunt procedure forrefractory ascites. NEnglJMed 1995; 332: 1192-7.

16 Jalan R, Redhead DN, Simpson KJ, Elton RA, Hayes PC.Transjugular intrahepatic portosystemic stent-shunt(TIPSS): long-term follow up. Q J Med 1994; 87:565-73.

17 PeltzerNY,RingEJ,LaBergeJM,HaskalZJ,RadosevichPM,Gordon RL. Treatment of Budd-Chiari syndrome withtransjugular intrahepatic portosystemic shunt. J VascIntervent Radiol 1993; 4: 263-7.

18 Strauss RM, Martin LG, Kaufman SL, Boyer TD. Trans-jugular intrahepatic portal systemic shunt for the manage-ment of symptomatic cirrhotic hydrothorax. Am JGastroenterol 1994; 89: 1520-2.

19 Richter GM, Noeledge G, Palmaz JC. The transjugularintrahepatic portosystemic stent-shunt (TIPSS): experi-ence and results of a pilot study. Cardiovasc InterventRadiol 1990; 13: 200-7.

20 Chalmers N, Redhead DN, Simpson KJ, Hayes PC. Trans-jugular intrahepatic portosystemic stent shunt (TIPSS):Early clinical experience. Clin Radiol 1992; 46: 166-9.

21 LaBerge JM, Somberg KA, Lake JR, et al. Two-year out-come following transjugular intrahepatic portosystemicshunt for variceal bleeding: results in 90 patients.Gastroenterology 1995; 108: 1143-51.

22 Rossle M, Haag K, Ochs A, et al. The transjugular intra-hepatic portosystemic stent-shunt procedure for varicealbleeding. NEnglJMed 1994; 330: 165-7 1.

23 Coldwell DM, Ring EJ, Rees CR, et al. MulticenterInvestigation of the role of transjugular intrahepaticportosystemic shunt in management of portal hyper-tension. Intervent Radiology 1995; 196: 335-40.

24 Franco D, Smadja C. Prevention of recurrent varicealbleeding: Surgical procedures. Clin Gastroenterol 1985; 12:233-57.

25 Matory WE Jr, Sedgwick CE, Rossi RL. Non-shuntingprocedures in management of bleeding oesophagealvarices. SurgClinNorthAm 1980; 60: 281-95.

26 Grace ND. The side-to-side shunt revisisted. N Engl JtMed1994; 330: 208-9.

27 Jalan R, Redhead DN, Ferguson J, Simpson KJ, Elton RA,Hayes PC. Wallstents or Palmaz stents for the tranajugularintrahepatic portosystemic stent-shunt (TIPSS). JfInteroent Radiol 1994; 9: 147-52.

28 Jalan R, Ferguson J, Allan PL, Redhead DN, Hayes PC.The role of duplex and colour flow doppler utrasonog-raphy in the assessment of TIPSS shunt function. Gut1995; 36 (suppl 1): A20.

29 Hou MC, Lin HC, Kuo BI, Chen CH, Lee FY, Lee SD.Comparison of endoscopic variceal injection sclero-therapy and ligation for the treatment of esophageal

on May 22, 2021 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.39.3.479 on 1 S

eptember 1996. D

ownloaded from

http://gut.bmj.com/

TIPSSfor the treatment ofportal hypertension 485

variceal hemorrhage: a prospective randomized trial.Hepatology 1995; 21: 1517-22.

30 Lacy AM, Nevasa M, Gilabert R, et al. Long-term affectsof distal splenorenal shunt on hepatic haemodynamicsand liver function in patients with cirrhosis: importanceof reversal of portal blood flow. Hepatology 1992; 15:616-22.

31 Johansen K. Prospective comparison of portal versus totalportal decompression for bleeding oesophageal varices.Surg Gynaecol obstet 1992; 175: 528-34.

32 Langer B, Taylor BR, Mackenzie DR, et al. Further reportof a prospective randomised trial comparing distalsplenorenal shunt with end-to-side portacaval shunt.Gastroenterology 1985; 88: 424-9.

33 Sanyal AJ, Freedman AM, Shiffman ML, Purdum PP,Leketic VA, Cheatham AK. Portosystemic encephalo-pathy after transjugular intrahepatic portosystemic shunt:results of a prospective controlled study. Hepatology 1994;20: 46-55.

34 Lake JR, Ring EJ, LaBerge J, Gordon R, Roberts J, AscherN.Transjugular intrahepatic portocaval stent shunts inpatients with renal insufficiency. TransplantProc 1993; 25:1766-7.

35 Martinet JP, Legault L, Cernacek P, et al. Effect of TIPSSon splanchnic and renal production of endothelins.Hepatology 1994; 20: 1 13A.

36 Rossle M, Sellinger M, Ochs A, Haag K. Liver function andblood flow before and after TIPSS. Hepatology 1993; 18:285A.

37 Crippin JS, Gross JB, Lindor KD. Increased intracranialpressure and hepatic encephalopathy in chronic liverdisease. Am J Gastroenterol 1992; 87: 879-82.

38 Henderson JM, Millikan WJ, Warren WD. The distalsplenorenal shunt: an update. World Jf Surg 1984; 8:722-32.

39 Jalan R, Elton R, Redhead DN, Finlayson NDC, Hayes PC.Analysis of prognostic variables in the prediction ofmortality, shunt failure, variceal rebleeding andencephalopathy following the transjugular intrahepaticportosystemic stent-shunt for variceal haemorrhage. JHepatol 1995; 23: 123-8.

40 John TG, Jalan R, Redhead DN, Hayes PC, Sanfey HA,Garden OJ. Transjugular intrahepatic portosystemicstent-shunt (TIPSS) insertion as a preclude to orthotopicliver transplantation (OLT) in patients with severe portalhypertension. Gut 1994; 35 (suppl 5): S24.

on May 22, 2021 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.39.3.479 on 1 S

eptember 1996. D

ownloaded from