Transjugular Intrahepatic Portosystemic Shunt — interventional decompression of the cirrhotic portal system
Transjugular Intrahepatic Portosystemic Shunt (TIPS) creates a low-resistance conduit between the portal and systemic venous circulations, decompressing a cirrhotic portal system without the mortality risk of open surgical shunting. Because TIPS diverts portal blood around the liver, appropriate patient selection — balancing the hemodynamic benefit against the risk of hepatic decompensation and encephalopathy — is the single most important determinant of outcome.
Refractory variceal bleeding is the best-established indication: gastroesophageal variceal hemorrhage that persists or recurs despite combined endoscopic therapy (band ligation, sclerotherapy) and pharmacologic control (octreotide, terlipressin, or vasopressin infusions plus non-selective beta-blockade). Early TIPS, placed within 24–72 hours of admission in high-risk patients (Child-Pugh C, or Child-Pugh B with active bleeding at endoscopy), reduces treatment failure and mortality compared with a rescue-only strategy.
Refractory ascites — fluid that fails to respond to maximal diuretic doses (spironolactone 400 mg/day plus furosemide 160 mg/day) or that recurs rapidly after paracentesis, or where diuretics are limited by renal impairment or electrolyte disturbance — is the second major indication. Patients requiring large-volume paracentesis (LVP) more than twice monthly are typical candidates.
Hepatic hydrothorax, a transdiaphragmatic pleural transudate occurring in 5–10% of cirrhotics with ascites, refractory to diuretics and sodium restriction, is a less common but well-validated indication, particularly when it is recurrent despite thoracentesis.
Severe or recurrent hepatic encephalopathy (grade ≥2, refractory to medical therapy) is a strong contraindication: TIPS mechanically bypasses the hepatocyte mass responsible for ammonia and toxin clearance, so patients already prone to encephalopathy are placed at unacceptable risk of debilitating or persistent confusion.
Severe congestive heart failure or moderate-to-severe pulmonary hypertension (mean pulmonary artery pressure >45 mmHg) is contraindicated because TIPS causes an abrupt increase in venous return and right heart preload — a circulation already limited by poor cardiac reserve can decompensate acutely.
Extensive or completely occlusive portal vein thrombosis historically precluded TIPS by removing the target for portal puncture, although modern techniques (transsplenic or transhepatic access, sharp PVT recanalization) have expanded candidacy in experienced centers; segmental, non-occlusive thrombus is not a contraindication.
Other relative or absolute contraindications include severe tricuspid regurgitation (transmits right heart pressure directly into the portal system), polycystic liver disease (cyst architecture obliterates safe puncture planes and risks hemorrhage), uncontrolled systemic infection or sepsis, and severe unremediated coagulopathy.
The Child-Pugh score (bilirubin, albumin, INR, ascites, encephalopathy — classes A, B, C) and the Model for End-Stage Liver Disease (MELD, using bilirubin, INR, creatinine ± sodium) are both used pre-procedurally to estimate hepatic reserve and post-TIPS mortality risk. Elective TIPS is generally favored in Child-Pugh A/B patients and MELD scores below roughly 18; MELD ≥18–24 confers substantially elevated 30- to 90-day mortality, and MELD >24 is often considered prohibitive outside of a bridge-to-transplant context. These scores do not replace clinical judgment — bilirubin trajectory, coagulopathy, and cardiac function are all weighed together — but they anchor the shared decision-making conversation about procedural risk versus the risk of continued bleeding or ascites.
A patient with Child-Pugh C cirrhosis and MELD 26 who is actively hemorrhaging still may require emergent TIPS as a life-saving measure despite an elevated MELD — the risk calculus in salvage bleeding differs fundamentally from elective ascites management, where the same MELD would argue strongly against the procedure.
TIPS is performed entirely percutaneously through the venous system, without a surgical incision into the abdomen. The interventional radiologist navigates a needle from the right internal jugular vein, through the right atrium and inferior vena cava, into a hepatic vein, and then directly through several centimeters of liver parenchyma into a portal vein branch — a blind, fluoroscopically guided pass that is the technical crux of the entire procedure.
The right internal jugular vein is punctured under real-time ultrasound guidance, minimizing risk to the adjacent carotid artery. A vascular sheath is placed and a hydrophilic guidewire and catheter are advanced under fluoroscopy through the superior vena cava, right atrium, and inferior vena cava into a hepatic vein — most commonly the right hepatic vein, which offers the most direct trajectory to the right portal vein, although the middle or left hepatic vein may be selected depending on portal venous anatomy.
Once the catheter is wedged in a peripheral hepatic vein branch, wedged hepatic venography is performed: contrast injected against the wedged catheter refluxes retrograde through the hepatic sinusoids to opacify the portal venous branches indirectly, giving the operator a roadmap of portal vein location and branching pattern before the blind needle pass. Direct portal venography (if a transhepatic or transsplenic access is used as an adjunct) may supplement this roadmap in technically difficult cases.
A rigid curved needle — the Colapinto needle, or the Rösch-Uchida transjugular liver access set — is advanced through the previously positioned catheter/sheath system, out of the hepatic vein, through several centimeters of hepatic parenchyma, and into a branch of the portal vein (right or left portal vein, chosen based on the wedge venogram roadmap and target anatomy). The needle trajectory is aimed anteromedially and caudally from the hepatic vein toward the expected location of the portal vein bifurcation.
This pass is performed largely blind, guided by fluoroscopic landmarks, anatomic experience, and sometimes intraprocedural ultrasound or CO2 portography. Because the needle traverses the liver capsule and parenchyma outside of any vessel for a short segment, the pass carries the primary risk of capsular perforation and hemoperitoneum if it exits the liver surface or punctures the hepatic artery or bile duct instead of the intended portal branch.
Once the needle is believed to be within the portal vein, a small test injection of contrast confirms position: opacification of the portal venous system (main portal vein, and ideally visualization of splenic and superior mesenteric venous confluence or varices) verifies successful cannulation, as opposed to hepatic artery (pulsatile, arterial-phase contrast) or bile duct (does not opacify a branching venous tree) puncture.
A guidewire is then passed through the needle into the main portal vein, over which the access sheath is advanced. Direct portal vein pressure is measured at this point and compared with right atrial or IVC pressure obtained earlier in the case — the difference is the baseline portosystemic pressure gradient (conceptually analogous to hepatic venous pressure gradient, HVPG), which anchors the pre-procedure severity assessment and the post-stent target.
With a wire securely spanning from the portal vein through the parenchymal tract into the hepatic vein and central venous system, the tract is mechanically enlarged and permanently scaffolded with a covered stent-graft — converting a temporary needle track into a durable, low-resistance shunt engineered to remain patent for years rather than days.
Before stent deployment, the needle tract through the liver parenchyma is dilated with an angioplasty balloon, typically to 6–8 mm diameter, creating a channel wide enough to accept the delivery system and pre-forming the tract so the stent-graft deploys smoothly and achieves its full diameter. This step also serves as a functional test: transient balloon occlusion across the tract, combined with contrast injection, confirms the tract trajectory cleanly bridges portal and hepatic venous systems without extravasation.
The dominant device in modern practice is the Viatorr TIPS Endoprosthesis (W. L. Gore), a self-expanding nitinol stent-graft with an ePTFE (expanded polytetrafluoroethylene) covering over its central and hepatic-vein-side segments, typically sized 8–10 mm in diameter. Earlier bare metal stents, once standard, have been largely abandoned because uncovered tracts develop intimal hyperplasia and occlude at high rates (primary patency as low as 25–50% at one year); covered stent-grafts reduced this dramatically by preventing bile and tissue ingrowth through the stent interstices.
The device is deployed across the entire tract — from the portal vein, through the liver parenchyma, into the hepatic vein and typically flush with or just into the IVC — under fluoroscopic guidance, unsheathing the self-expanding nitinol frame which apposes against the tract walls and reaches nominal diameter over several minutes to hours.
A refinement of the Viatorr device — the controlled expansion (CX) Viatorr — allows the operator to under-dilate the stent initially (to a smaller diameter such as 8 mm) and, if the post-deployment portosystemic gradient remains above target, dilate further up to 10 mm without needing a second device, allowing graded titration of shunt flow to the individual patient's hemodynamic response.
Critically, the portal-vein-side segment of the device is left uncovered (bare nitinol) for roughly 2 cm. This design preserves some antegrade flow into adjacent intrahepatic portal branches rather than diverting all portal inflow through the shunt, which helps maintain a degree of nutrient- and hormone-rich portal perfusion to the hepatocytes and may reduce the incidence of post-TIPS hepatic dysfunction and encephalopathy compared with older fully covered or bare designs.
Stent diameter is a direct hemodynamic lever: an 8 mm shunt trades a smaller pressure-gradient reduction for lower encephalopathy risk, while a 10 mm shunt more reliably reaches the <12 mmHg target but shunts a larger fraction of portal flow around the liver, raising the risk of overshunting and hepatic encephalopathy. Modern controlled-expansion devices let the operator titrate diameter intraprocedurally against the measured gradient rather than committing to a fixed size upfront.
The technical endpoint of TIPS is a quantifiable one: a measured reduction in the portosystemic pressure gradient across the newly placed shunt. This single number predicts, better than any other intraprocedural measurement, whether variceal bleeding will be controlled and whether ascites will resolve — and it is checked before the patient ever leaves the angiography suite.
The portosystemic pressure gradient (PSG) is the direct difference between portal vein pressure — measured with a catheter placed in the main portal vein — and right atrial or inferior vena cava pressure measured at the same session. This is conceptually the direct-measurement analog of the hepatic venous pressure gradient (HVPG, wedged minus free hepatic venous pressure) used in non-interventional risk stratification, but obtained here by direct cannulation rather than balloon-occlusion.
Baseline PSG in patients undergoing TIPS for variceal bleeding or refractory ascites is typically well above the 10–12 mmHg threshold associated with clinically significant portal hypertension, often ranging from the high teens into the 20s or 30s mmHg in severe cirrhosis. The gradient is remeasured immediately after stent-graft deployment, with the catheter repositioned across the shunt.
The accepted procedural target is a final PSG below 12 mmHg, or — when baseline pressure is extremely elevated and reaching 12 mmHg would require an unsafely large shunt — a reduction of at least 50% from the pre-procedure baseline. Achieving this target correlates strongly with durable control of variceal hemorrhage: more than 90% of patients undergoing TIPS for refractory or high-risk variceal bleeding achieve immediate hemostasis, and rebleeding rates over the following year are substantially lower than with endoscopic therapy alone.
For refractory ascites, the relationship between gradient reduction and clinical response is present but less absolute — roughly 60–75% of patients experience resolution or marked improvement in ascites (freedom from large-volume paracentesis), reflecting that ascites formation depends not only on portal pressure but also on sodium handling, albumin, and systemic hemodynamics that TIPS influences but does not fully normalize.
Because TIPS acutely redirects a large volume of splanchnic blood flow into the systemic circulation, cardiac preload and cardiac output rise immediately post-procedure — typically a 15–30% increase in cardiac index — which is generally well tolerated in patients with normal cardiac reserve but is precisely why severe heart failure and pulmonary hypertension are screened for and excluded beforehand. Systemic vascular resistance falls as the effective arterial underfilling of advanced cirrhosis is partially corrected, and renal perfusion often improves, which underlies TIPS' efficacy against hepatorenal physiology-driven ascites as well as pure mechanical portal hypertension.
Successfully decompressing the portal system does not end the clinical relationship — TIPS converts an acute bleeding or ascites problem into a chronic shunt that must be monitored indefinitely. The two dominant long-term concerns are hepatic encephalopathy, a direct consequence of bypassing hepatic detoxification, and shunt stenosis or thrombosis, which silently reverses the hemodynamic benefit if undetected.
New or worsening hepatic encephalopathy is the most common clinically significant complication of TIPS, occurring in roughly 25–35% of patients, most often within the first weeks to months after shunt placement. The mechanism is direct: portal blood — carrying ammonia and other gut-derived nitrogenous compounds — is diverted around the hepatocyte mass responsible for their metabolism and detoxification, delivering a higher toxin load directly into the systemic circulation and across the blood-brain barrier.
Risk factors for post-TIPS encephalopathy include prior episodes of HE, age over 65, low serum albumin, sarcopenia, and larger shunt diameter (more complete portosystemic diversion). First-line management is medical: lactulose to acidify the colon and trap ammonia as non-absorbable ammonium, and rifaximin, a non-absorbed antibiotic that reduces ammonia-producing gut flora. Refractory or debilitating encephalopathy despite medical therapy may require shunt reduction — placement of a smaller-diameter stent or a constraining device within the existing stent-graft to partially restore hepatic portal perfusion — as a targeted, reversible intervention.
Because HE risk rises with age, low albumin, prior HE episodes, and shunt diameter, pre-procedure risk stratification lets the operator choose a smaller controlled-expansion stent diameter (8 mm rather than 10 mm) in high-risk patients — deliberately accepting a smaller gradient reduction in exchange for materially lower encephalopathy risk.
Even covered stent-grafts are subject to stenosis (typically at the hepatic venous end, from pseudointimal hyperplasia) or thrombosis, both of which silently return the patient to a pre-TIPS hemodynamic state and reintroduce bleeding or ascites risk. Surveillance duplex Doppler ultrasound is performed on a structured schedule — commonly at 1 week, 1 month, 3 months, 6 months, and 12 months post-procedure, then periodically thereafter — measuring peak shunt velocity, directionality of flow, and portal vein velocity.
A normal, widely patent covered stent typically shows shunt velocities in the range of roughly 90–190 cm/s with maintained hepatopetal-to-shunt flow; a significant fall in velocity, reversal of flow direction, or velocity outside the expected range prompts venography and, if a stenosis is confirmed, balloon angioplasty or relining with an additional stent segment to restore patency. This surveillance-driven, minimally invasive maintenance is a major reason covered stent-grafts achieve primary patency of roughly 80–90% at one year, versus historically 25–50% for uncovered bare-metal stents.
Hemolysis — typically mild and self-limited — occurs in a subset of patients due to mechanical shear stress on red cells passing through the stent-graft. Right heart strain can occur in patients with marginal cardiac reserve who receive an unexpectedly large preload increase; this is largely mitigated by pre-procedure cardiac screening. Capsular perforation with hemoperitoneum, historically one of the most feared complications of the blind transhepatic needle pass, has become rare with modern covered stent-grafts, ultrasound-assisted planning, and operator experience, though it remains a recognized risk of the access phase rather than the deployment phase. Procedural or post-procedural infection, including infection of the stent-graft itself, is uncommon but is treated aggressively given the difficulty of removing an intravascular device once endothelialized.