Early TIPS

Published

García-Pagán JC, et al. "Early use of TIPS in patients with cirrhosis and variceal bleeding". The New England Journal of Medicine. 2010. 362(25):2370-2379.
PubMed • Full text • PDF

Clinical Question

Among selected high-risk patients with cirrhosis hospitalized for acute variceal bleeding, does the early use of transjugular intrahepatic portosystemic shunt (TIPS) reduce the treatment failure and mortality rate?

Bottom Line

Among selected high-risk patients with hepatic cirrhosis hospitalized for acute variceal bleeding, early TIPS improves one-year overall survival and reduces the risk of rebleeding or failure to control bleeding when compared to pharmacotherapy group.

Major Points

Patients with hepatic cirrhosis and portal hypertension develop portosystemic collaterals including gastroesophageal varices, rupture of which results in variceal hemorrhage. Meta-analyses have demonstrated that rapid pharmacologic and endoscopic management of variceal bleeding leads to early resolution of hemorrhage and improves hemodynamic parameters, but does not necessarily improve survival. Failure rates remain high, and mortality may be as high as 20%. In patients with refractory variceal hemorrhage, transjugular intrahepatic portosystemic shunt (TIPS) placement may reduce the portosystemic pressure gradient and reduce bleeding from active varices. One study demonstrated the efficacy of early TIPS in patients with variceal hemorrhage refractory to pharmacologic and endoscopic interventions, but this study was limited by less aggressive therapy administered to the control group, and by the use of older (non-contemporary) uncovered TIPS stents. Additionally some studies suggest that early TIPS reduces rates of variceal bleeding without improving survival. Therefore, a properly controlled randomized trial of early TIPS was required to more conclusively demonstrate its efficacy.

García-Pagán and colleagues conducted this randomized controlled trial in which 63 patients with cirrhosis and acute variceal hemorrhage with high-risk features were randomized to either early TIPS or pharmacologic plus endoscopic therapy. During the follow-up of 16 months, the incidence of rebleeding or failure to control bleeding was higher in the pharmacotherapy/endoscopy group as compared to the early TIPS group (14 vs. 1 failures; P=0.001). The 1-year actuarial probability of freedom from rebleeding was lower in the pharmacotherapy/endoscopy group than the early TIPS group (50% vs. 97%; P<0.001). In addition, the mortality rate was significant higher in the pharmacotherapy/endoscopy group (12 vs. 4; P=0.01). Rates of hepatic encephalopathy were similar between groups.

Given the stringent eligibility requirements for this study, selection is key if early TIPS is being considered for an individual patient. This is in line with various guideline statements including those from the AASLD.

Guidelines

AASLD Cirrhotic Variceal Hemorrhage (2016, adapted):

• In patients at high risk of failure or rebleeding (CTP class C cirrhosis or CTP class B with active bleeding on endoscopy) who have no contraindications for TIPS, an “early” (pre-emptive) TIPS within 72 hours from EGD/EVL may benefit selected patients.
• For patients in whom an early TIPS is not performed, intravenous vasoactive drugs should be continued for 2-5 days and NSBBs initiated once vasoactive drugs are discontinued. Rescue TIPS is indicated in these patients if hemorrhage cannot be controlled or if bleeding recurs despite vasoactive drugs.
• In patients in whom TIPS is performed successfully, intravenous vasoactive drugs can be discontinued.

Design

• Multicenter, double-blind, randomized, controlled trial
• N=63 patients with high-risk cirrhotic variceal hemorrhage
  • Pharmacotherapy/endoscopic therapy (n=31)
  • Early TIPS (n=32)
• Setting: 9 European centers
• Enrollment: 2004-2007
• Follow-up: median 16 months
• Analysis: Intention-to-treat
• Primary outcome: freedom from uncontrolled bleeding or rebleeding within 1 year

Population

Inclusion Criteria

• Cirrhosis with acute esophageal variceal bleeding
• Currently treated with a combination of vasoactive drugs, endoscopic treatment and prophylactic antibiotics.
• Child-Pugh class C (score 10-13) or Child-Pugh class B (score 7-9) with active bleeding at diagnostic endoscopy

Exclusion Criteria

• Age >75 years
• Child-Pugh score >13
• Pregnancy
• Hepatocellular carcinoma that did not meet the Milano criteria for transplantation
• Creatinine level >3 mg/dL
• Previous pharmacologic therapy combined endoscopic treatment to prevent rebleeding
• Previous use of a portosystemic shunt or TIPS
• Bleeding from isolated gastric or ectopic varices
• Total portal-vein thrombosis
• Heart failure

Baseline Characteristics

From the pharmacotherapy/endoscopy group.

• Age, mean: 49 years
• Female: 25%
• Cause of liver cirrhosis: alcohol (65%), HCV (16%), other (19%)
• Active alcoholism (no. of patients): 17 vs. 15 (P = 0.61)
• Grading of liver dysfunction
  • Child-Pugh Class B (no. of patients): 16 vs. 16 (P = 0.99)
  • Child-Pugh Class C (no. of patients): 15 vs. 16 (P = 0.99)
  • Child-Pugh score: 9.5±1.8 vs. 9.3±1.8 (P = 0.66)
  • MELD score: 16.9±6.3 vs. 15.5±5 (P = 0.28)
  • MELD-Na score: 19±7 vs 17±6 (P = 0.25)
• Ascites (no. of patients): 18 vs. 19 (P = 0.99)
• Baseline laboratory results
  • Bilirubin (mg/dL): 4.4±4.9 vs. 3.7±4.8 (P = 0.34)
  • Albumin (g/L): 26±7 vs. 26±7 (P = 0.89)
  • Prothrombin time (%): 50±15 vs. 53±15 (P = 0.33)
  • Creatinine (mg/dL): 1±0.4 vs. 1±0.5 (P = 0.59)
• Previous hepatic encephalopathy (no. of patients): 0 vs. 6 (P = 0.03)
• Previous variceal bleeding in the absence of combined treatment (no. of patients): 5 vs. 11 (P = 0.15)
• Patient's condition and interventions at admission
  • Active bleeding at endoscopy (no. of patients): 22 vs. 23 (P = 0.99)
  • Shock at time of admission (no. of patients): 7 vs. 7 (P = 0.99)
  • Systolic BP (mmHg): 111±23 vs. 104±18 (P = 0.22)
  • Haematocrit on admission (%): 27±8 vs. 26±6 (P = 0.49)
  • Units of blood transfused before randomisation (no.): 2.9±3 vs. 2.7±2 (P = 0.73)
  • Endoscopic treatment at time of index bleeding (no. of patients) (P = 0.17)
    • Band ligation: 26 vs. 21
    • Injection sclerotherapy: 5 vs. 11
  • Vasoactive-drug therapy at time of index bleeding (no. of patients) (P = 0.74)
    • Telipressin: 14 vs. 12
    • Somatostatin: 16 vs. 18
    • Octreotide: 1 vs. 2

Interventions

• 63 patients were randomised independently within 24 hours after admission with a computer-generated randomisation sequence.
  • 31 patients were assigned to pharmacotherapy and endoscopic band ligation (EBL).
  • 32 patients were assigned to receive early TIPS.
• Pharmacotherapy plus EBL
  • Initially treated with vasoactive drugs until free of bleeding for at least 24 hours and preferably up to 5 days.
  • Subsequently commenced non-selective beta-blocker (either propranolol or nadolol) with dose increased in step-wise fashion every 2 - 3 days to maximum tolerated dose.
    • 22 patients received propranolol and only 3 received nadolol.
    • Beta-blocker therapy was not commenced for the remaining 6 patients due to failure to control bleeding, early rebleeding, or death.
  • 10 mg ISMN was commenced at bedtime and increased in stepwise manner to the maximum tolerated dose.
    • ISMN was added to 12 patients' drug regimen.
    • ISMN was not commenced for 13 patients due to arterial hypotension, treating physician's preference or early death.
  • Within 7 - 14 days, second EBL was performed and scheduled every 10 - 14 days until varices were eradicated.
  • Received PPI until variceal eradication was accomplished.
  • Endoscopic monitoring at 1-month, 6-month and 12-month intervals and then annually.
• Early TIPS
  • Vasoactive drugs were administered until TIPS was performed within 72 hours after diagnostic endoscopy.
  • With e-PTFE-covered stents initially dilated to 8 mm and further dilated to 10 mm if portal pressure gradient did not fall to < 12 mmHg.
• Follow up visit
  • Scheduled at 1 month, 3 month and every 3 month thereafter.
  • Doppler ultrasound at first visit, 6 month and every 6 month thereafter.
  • Followed until death or liver transplantation (maximum of 2 years or until end of the study).

Outcomes

Comparisons are pharmacotherapy/endoscopy group vs. early TIPS.

Primary Outcomes

Failure to control acute bleeding or rebleeding within 1 year

45.2% vs. 3.1% (P=0.001; NNT 2)

Secondary Outcomes

Treatment failure

At 5 days: 4/31 vs. 1/32 patients

> 5 days - 6 weeks: 7/31 vs. 0/32 patients

> 6 weeks - 1 year: 3/31 vs. 0/32 patients

Minor rebleeding

Total no. of cases: 3/31 vs. 1/32 patients (P=0.35)

Gastric ulcer: 1/31 vs. 0/32 patients

Portal hypertensive gastropathy: 1/31 vs. 0/32 patients

Varices: 1/31 vs. 0/32 patients

Post-EBL ulcer: 0/31 vs. 1/32 patients

1-year actuarial probability of freedom from uncontrolled bleeding or rebleeding

50% vs. 97% (ARR: 47%; 95% CI 25 to 69; NNT: 2.1; 95% CI 1.4 to 4.0; P=0.001)

Death

12/31 vs. 4/32 patients (P=0.01)

Child-Pugh Class B: 2/31 vs. 1/32 patients

Child-Pugh Class C: 10/31 vs. 3/32 patients

Cause of death

Recurrent bleeding: 5/31 vs. 0/32 patients

Sepsis: 4/31 vs. 3/32 patients

Liver failure: 2/31 vs. 0/32 patients

Hepatorenal syndrome: 1/31 vs. 0/32 patients

Arrhythmia: 0/31 vs. 1/32 patients

6-week survival rate

67% vs. 97% (ARR: 30%; 95% CI 12 to 48; NNT: 3.3; 95% CI 2.1 to 8.3; P=0.001)

1-year survival rate

61% vs. 86% (ARR: 25%; 95% CI 2 to 48; NNT: 4; 95% CI 2.1 to 50; P=0.001)

Days in intensive care unit (no.)

8.6 vs. 3.6 (P=0.01)

Orthotopic liver transplantation

2/31 vs. 4/32 patients (P=0.67)

Complications of portal hypertension

Hepatic encephalopathy: 12/31 vs. 8/32 patients

Ascites: 9/31 vs. 5/32 patients

Spontaneous bacterial peritonitis: 2/31 vs. 0/32 patients

Hepatorenal syndrome: 5/31 vs. 2/32 patients

Hepatic encephalopathy at 1 year

40% vs. 28% (95% CI -18-40%; P=0.13)

New or worsening ascites at 1 year

33% vs. 13% (95% CI -8 to 47%; P=0.11)

Adverse Events

Other serious adverse events (no. of patients)

Infections (other than pneumonia or SBP): 4/31 vs. 4/32

Pneumonia: 3/31 vs. 3/32

Acute episode in chronic liver failure: 1/31 vs. 3/32

Delirium tremens: 2/31 vs. 1/32

Korsakoff syndrome: 0/31 vs. 1/32

Oesophageal perforation: 1/31 vs. 0/32

Dysphagia: 1/31 vs. 0/32

Bleeding from banding-related ulcers: 1/31 vs. 0/32

Choledocholithiasis: 0/31 vs. 1/32

Alcoholic hepatitis: 0/31 vs. 1/32

Bleeding duodenal ulcer: 0/31 vs. 1/32

Non-serious adverse events (no. of patients)

Gynaecomastia: 0/31 vs. 1/32

Oedema: 1/31 vs. 0/32

Sleep disorder: 1/31 vs. 0/32

Chest pain (after EBL): 1/31 vs. 0/32

Epistaxis (after EBL): 1/31 vs. 0/32

Vomiting: 1/31 vs. 0/32

Criticisms

From the New England Journal of Medicine correspondence:

• The study did not explain how cirrhosis was diagnosed.
• Co-existing alcoholic liver disease could lead to the possibility of overestimating disease severity.
• Overestimating disease severity could affect the timing of calculation of prognostic scores and account for the improved survival in patients with either Child-Pugh Class B or C disease.
• There is selection bias in the study because patients presenting with encephalopathy were not included.
• Patients with previous hepatic encephalopathy, as a potential confounder in the study, were overrepresented in the early-TIPS group that may have interfered with their conclusion about rate of hepatic encephalopathy during index bleeding and 1-year follow-up.

Funding

Funding comes from Instituto de Salud Carlos III, Spain (FIS 06/0623, 04/1031 and 08/0504), Ministerio de Educación y Ciencia, Spain (SAF 07/61298), University Hospital Center of Toulouse, Fund for Scientific Research — Flanders (Fundamenteel Klinisch Mandaat — FWO Vlaanderen, to Dr. Laleman) and an educational grant from Gore.