Emergent Transjugular Intrahepatic Portosystemic Shunt Creation in Acute Variceal Bleeding

Abstract

Emergent transjugular intrahepatic portosystemic shunt (TIPS) creation is most commonly employed in the setting of acute variceal hemorrhage. Given a propensity for decompensation, these patients often require a multidisciplinary, multimodal approach involving prompt diagnosis, pharmacologic therapy, and endoscopic intervention. While successful in the majority of cases, failure to medically control initial bleeding can prompt interventional radiology consultation for emergent portal decompression via TIPS creation. This article discusses TIPS creation in emergent, acute variceal hemorrhage, reviewing the natural history of gastroesophageal varices, presentation and diagnosis of acute variceal hemorrhage, pharmacologic therapy, endoscopic approaches, patient selection and risk stratification for TIPS, technical considerations for TIPS creation, adjunctive embolotherapy, and the role of salvage TIPS versus early TIPS in acute variceal hemorrhage.

Emergent transjugular intrahepatic portosystemic shunt (TIPS) creation is most commonly employed in the setting of acute variceal hemorrhage (VH), a dreaded complication of portal hypertension in patients with liver cirrhosis. Given a propensity for decompensation, these patients often require a multidisciplinary, multimodal approach involving prompt diagnosis, pharmacologic therapy, and endoscopic intervention. ^{1 2 3 4} While successful in the majority of cases, failure to medically control initial bleeding can prompt interventional radiology (IR) consultation for emergent portal decompression via TIPS creation. This article discusses TIPS creation in emergent, acute VH, reviewing the natural history of gastroesophageal varices (GEVs), presentation and diagnosis of acute VH, pharmacologic therapy, endoscopic approaches, patient selection and risk stratification for TIPS, technical considerations for TIPS creation, adjunctive embolotherapy, and the role of salvage TIPS versus early TIPS in acute VH.

Gastroesophageal Varices

Portal hypertensive hepatic venous pressure gradients (HPVGs)—defined as pressure gradients ≥ 10 mm Hg—result in GEV formation via a dynamic, multifactorial process. ^{3 5} GEVs progress with worsening liver cirrhosis, and are categorized as “small” (< 5 mm), “medium” (based on tortuous morphology and occupation of less than one-third of the esophageal lumen), and “large” (> 5 mm). ^{3 6 7 8} Classically, EVs are reported in 52% of endoscopically screened patients, and increase in prevalence with disease severity, conservatively ranging from 43% in patients with Child-Pugh class A disease to 72% of patients with Child-Pugh class B/C disease. ⁹ EVs are reported to have an annual formation incidence of 7 to 9% ^{10 11 12} and an annual progression rate of 10 to 15%. ^{13 14 15} Given these rates, approximately 90% of cirrhotic patients have EVs at 10 years. ¹⁶ In contrast, GVs are less prevalent, and are reported in 5 to 33% of patients with liver cirrhosis. ^{17 18} Today, EVs and GVs are delineated as unique anatomic entities with varying physiologic behavior. EVs exist at higher pressures than GV: 16 to 24 mm Hg vs. 11 to 18 mm Hg. ^{13 14} While problematic EVs are historically located within the lower 5 cm of the distal esophagus, ^{19 20} GVs exist at locations which are demarcated by the Sarin classification system. ¹⁷ In this classification scheme, gastroesophageal varices type 1 (GEV1) occur at the lesser curvature of the stomach, gastroesophageal varices type 2 (GEV2) at the greater curvature of the stomach, and isolated gastric varices type 1 (IGV1) at the fundal region. Ectopic isolated gastric varices type 2 (IGV2) are rarely encountered.

EV hemorrhage is generally thought to account for up to 70% of upper gastrointestinal (GI) bleeding in patients with liver cirrhosis, ^{1 21} and occurs at variable rates. Small EV hemorrhage is known to occur at 5% yearly, versus 15% annually for medium and large EV hemorrhage. ²² Comparatively, GV hemorrhage occurs 50% less frequently than EV hemorrhage, and is attributed to between 10 and 30% of all VH. ¹⁷ Although less frequent, GV hemorrhage rates are reported to occur at 16% at 1 year, 36% at 3 years, and 44% at 5 years, often with high transfusion requirements due to severity of bleeding. ^{17 18} Data show that GV hemorrhage occurs at lower portosystemic gradients (PSGs)—approximating 16 to 20 mm Hg—when compared with EV hemorrhage (21–23 mm Hg), with 10% of GV hemorrhage and 2% of EV hemorrhage occurring at PSGs below 12 mm Hg. ^{23 24}

Prevention of First Variceal Hemorrhage

Given the high prevalence of varices in liver cirrhosis, patients with this condition undergo screening endoscopy at the time of liver disease diagnosis, with follow-up surveillance every 2 to 3 years, or every 1 to 2 years if small varices are present. ^{3 25 26} Additionally, endoscopy is recommended upon declaration of cirrhotic decompensation and yearly thereafter. ^{3 7 26} Ideally, this strategy allows for prevention of first VH.

Prevention of VH is best understood in the setting of EV. Treatment hinges on maintaining a protective HPVG ≤ 12 mm Hg, ^{3 27} with definitive endoscopic treatment of varices. Patients with medium and large EVs, and those with decompensated cirrhosis or stigmata of bleeding—such as red wale signs—are often treated with nonselective β-blockade (using propranolol, nadolol, or carvedilol) or endoscopic variceal ligation (EVL) to potentially prevent variceal progression and hemorrhage. ^{3 25 28 29} Combination therapy is usually not performed. ³

Comparatively, prevention of GV hemorrhage is less understood. A randomized controlled trial (RCT) examined the use of sclerosant glue compared with nonselective β-blockade or observation and demonstrated similar survival between β-blockade and sclerosant utilization. ^{3 30} Prospective evaluation of variceal obliteration is yet to be adequately performed in this setting. ³¹

Diagnosis of Variceal Hemorrhage

Patients with acute VH often present with hematemesis or melena, and variable states of hemodynamic stability. Assessment begins by obtaining a detailed history from the patient or family, complemented by thorough chart and imaging review, which will often yield a history of liver cirrhosis, portal hypertension, and/or GEVs. Baseline laboratory assessment includes a complete blood count, comprehensive metabolic panel, and a basic coagulation panel.

Differentiation between acute VH and other sources of upper GI bleeding in the cirrhotic patient is an important consideration, as this population experiences nontrivial rates of peptic ulcer disease. ^{32 33} As the interventional approach to arterial and variceal bleeding sources differs substantially, clear delineation of the hemorrhage source is compulsory ( Fig. 1 ). While endoscopy serves as the primary means to diagnose bleeding sources, multiphase computed tomography angiography (CTA) may serve as an adjunctive modality to determine whether there is active arterial hemorrhage, determine a source, and provide a potential roadmap for intervention. ³⁴ Additionally, a properly performed venous phase can help identify ectopic (e.g., small bowel) varices that cannot be identified with conventional endoscopy. Intravenous proton pump inhibitors (PPIs) are often initiated in the setting of upper GI bleeding, and their use remains controversial in cirrhotic patients. ¹ PPIs may be discontinued upon confirmation of VH ⁴ given an association with spontaneous bacterial peritonitis. ³⁵

Fig. 1.

TIPS in the setting of elevated MELD score. A 49-year-old woman with hepatitis C virus liver cirrhosis referred to IR for emergent TIPS in the setting of massive EV bleeding and disseminated intravascular coagulation. At the time of referral, patient failed endoscopic treatment and was in critical condition with calculated MELD score of 40. TIPS created uneventfully ( arrow ), and laboratory and clinical findings following procedure supported bleeding cessation. Nonetheless, patient's condition did not improve, a do-not resuscitate status was assigned, and she died due to multiorgan failure 2 days post-TIPS. Though TIPS may be applied in acutely bleeding patients with high MELD scores, likelihood of survival is extremely poor. Given poor anticipated clinical outcomes, open discussion that conveys anticipated course to patient and family is critical in clinical decision-making process.

Pharmacotherapy, Endoscopy, and Bridge to Salvage TIPS

Upon confirmation of acute VH, treatment aims to achieve hemostasis, prevent early rebleeding, and minimize 6-week mortality. ^{3 25 36} Patients experiencing acute VH demand high acuity of care, often requiring monitoring within an intensive care unit equipped with skilled nursing staff and a multidisciplinary team of critical care specialists, hepatologists, GI endoscopists, interventional radiologists (IRs), and surgeons. ²⁶

Initial management involves airway control, especially in the setting of hematemesis and hepatic encephalopathy (HE). Prophylactic intubation is controversial ²⁶ and has been associated with higher risk of aspiration pneumonia. ³⁷ Thus, the decision to intubate is made on a case-by-case basis.

Volume resuscitation is challenging, and conservative transfusion protocols are utilized. A mean arterial pressure goal of approximately 65 mm Hg—extrapolated from data regarding the management of septic shock and trauma ^{38 39 40 41} —is usually pursued. ¹ Hypovolemia at presentation may temper VH by decreasing portal pressure, while aggressive resuscitation can aggravate VH or promote rebleeding by elevating portal pressure. ^{4 42 43} A RCT demonstrated improved survival in patients with Child-Pugh class A and B disease, with decreased overall rebleeding in all Child-Pugh class A, B, and C patients, when transfusion occurred below hemoglobin of 7 g/dL with a posttransfusion target of 7 to 9 g/dL. ^{4 44} Current treatment recommendations incorporate these parameters. ³

Pharmacologic therapy to achieve hemostasis emphasizes splanchnic vasoconstriction in lieu of correction of coagulopathy. No benefit has been shown when factor VII is administered to correct prothrombin time. ^{4 45 46} Moreover, no compelling data are available regarding correction of thrombocytopenia or use of desmopressin. ^{3 4 47} Current pharmacologic options to achieve hemostasis include intravenous splanchnic vasoconstriction with octreotide, somatostatin, or terlipressin, administered for up to 3 to 5 days. ^{3 25 48} This recommendation is supported by a meta-analysis of RCTs that demonstrated improved hemostasis, decreased blood transfusions, and lower 7-day mortality with utilization of splanchnic vasoconstriction. ^{3 49}

Antibiotic prophylaxis is initiated, given high risk for cirrhotic patients with upper GI bleeding to develop gram-negative bacterial infections, ^{1 50 51 52} which are postulated to occur due to translocation of enteral flora. RCTs and meta-analyses exploring antibiotic prophylaxis have demonstrated decreased infection rates, rebleeding, and mortality. ^{3 50 52} Due to prevalent quinolone resistance, 1 g ceftriaxone is usually administered every 24 hours for 7 days. ^{3 25 53}

Prompt endoscopy (within 12 hours) is recommended, ^{3 25} as patients experiencing VH may rapidly decompensate. Endoscopy allows for definitive treatment of bleeding varices by EVL or endoscopic injection sclerotherapy (EIS). ^{26 31 54 55} Data show increased in-hospital mortality when endoscopy is delayed by more than 15 hours in hemodynamically stable patients, ^{4 56} and that endoscopy within 12 hours is associated with lower rates of rebleeding and mortality. ⁵⁷ Timing remains debated, as others report no overall impact on mortality with delayed endoscopy in stable patients. ^{56 58 59}

While EV hemorrhage is typically treated by combination EVL and splanchnic vasoconstriction, ^{3 4 25 48} GV hemorrhage is often difficult to ameliorate by EVL secondary to anatomic considerations, and suboptimal EVL can lead to ulceration and subsequent bleeding. ^{31 60} While a meta-analysis has shown that cyanoacrylate glue is associated with lower rebleeding rates in GVs, EVL and EIS have comparable rates of hemostasis. ^{3 61}

Despite attempts to optimize combined pharmacotherapy and endoscopic treatment, acute VH is refractory in approximately 10 to 20% of patients, ^{16 62} and temporizing measures must be considered prior to rapid portal decompression via TIPS creation. Today, bridging measures such as self-expanding metal stent (SEMS) deployment or 24-hour balloon tamponade (using Sengstaken-Blakemore, Linton-Nachlas, or Minnesota tubes) are utilized. Overall, utilization of balloon tamponade correlates with severity of bleeding and is successful in up to 80% of patients, although its use in VH is associated with increased adverse events and mortality rates of 20%. ^{3 7 63 64} Comparatively, SEMS can be left in place for up to 7 days ⁴ and have demonstrated both better hemostasis and less adverse events, ^{3 65 66 67} although stent migration remains a worrisome complication. ⁶⁸

TIPS Indications

Current indications for TIPS creation in VH are summarized by the American Association for the Study of Liver Diseases (AASLD) ³ :

1. Rescue therapy for acute VH which is refractory to pharmacologic and endoscopic treatment.

2. Recurrent VH despite optimization of pharmacology and endoscopic treatment.

3. Emerging indication for early (within 24–72 hours) TIPS creation in acute VH.

TIPS is not pursued for prevention of VH in nonbleeding varices, prevention of rebleeding in patients who have experienced only one episode of EV hemorrhage, prevention of hemorrhage from gastric antral vascular ectasia, or portal hypertensive gastropathy. ^{69 70}

While no absolute contraindications exist, relative contraindications for TIPS creation are presented by the Society for Interventional Radiology. ⁷¹ Briefly, relative contraindications include heart failure or major valvular insufficiency, elevated left or right heart pressures, rapidly progressive liver failure, severe or uncontrolled HE, uncontrolled systemic infection, sepsis, unrelieved biliary obstruction, polycystic liver disease, extensive primary or metastatic hepatic malignancy, and severe uncontrolled coagulopathy. ⁷¹

Patient Selection and Risk Stratification

Patient selection is a challenging endeavor for the IR, as individuals experiencing acute VH often have decompensated cirrhosis. It has long been known that the markedly elevated portal pressures in decompensated cirrhosis contribute to morbidity and mortality in VH. HPVG measurements ≥ 20 mm Hg are associated with failure to achieve hemostasis and early rebleeding. ^{72 73} Moreover, HPVGs ≥ 16 mm Hg have also been associated with mortality ²² and elevated serum bilirubin has been reported to increase risk of mortality with each 1 mg/dL increase above a threshold of 3.0 mg/dL. ⁷⁴ These and other objective surrogate markers guide decision making.

Currently, the Model for End-Stage Liver Disease (MELD) score is the predominant system utilized for objective risk stratification and patient selection for TIPS creation. ⁷⁵ MELD utilizes serum total bilirubin, international normalized ratio (INR), and serum creatinine as surrogate markers to assess liver function and has classically and contemporaneously (with and without modification) demonstrated efficacy in predicting mortality. ^{76 77 78 79 80 81 82} Moderately elevated short-term mortality rates are evident in patients with MELD scores exceeding 18, and markedly elevated short-term mortality rates occur above the MELD threshold of 25 ( Fig. 2 ). ⁸³ Moreover, the MELD score has successfully been extrapolated in the emergent setting to predict 90-day mortality with high accuracy (area under the receiver operating characteristics: 0.842, 95% confidence interval [CI]: 0.755–0.928). ⁸⁴ Similarly, Conejo et al examined Child-Pugh and MELD scores in patients with decompensated cirrhosis experiencing acute VH, and reported that MELD scores exceeding 19 and Child-Pugh class C disease confer high risk of death. ⁸⁵

Fig. 2.

TIPS for presumed variceal bleeding with subsequent discovery of arterial bleeding source. A 73-year-old man with alcoholic cirrhosis and acute VH transferred from outside hospital for possible TIPS creation. Upper endoscopy performed 1 week prior revealed bleeding EVs. Upon transfer, recurrent EV bleeding presumed, though active bleeding source not established via repeat endoscopy. While TIPS was successfully created ( a ; arrow ), patient continued to have hematochezia and decreasing hemoglobin levels. Repeat upper endoscopy revealed bleeding duodenal ulcer. Arteriography confirmed active hemorrhage from gastroduodenal artery ( b ; arrow ), and bleeding cessation achieved after coil embolization ( c ; arrow ). Appropriate preprocedural diagnosis and workup is critical in establishing source of upper GI bleeding prior to therapy prescription, to recommend most suitable management approach.

Even with high predictive value, criteria such as MELD score cannot be solely utilized for patient selection. Despite poor prognosis in severely decompensated liver disease, emergent portal decompression by TIPS creation is considered a potentially lifesaving measure in refractory acute VH. Thus, patients should be selected on a case-by-case basis. Critical conversations, including a potential need for liver transplantation, are necessary between the patient (or family), referring service, and operating IR. Minor complication rates (up to 4%) should be discussed, along with major complication rates (up to 3%), and relevant potential risks including persistent VH (i.e., therapeutic failure), hepatic infarction, severe or life-altering HE, and death. ⁷¹ Additionally, if GV hemorrhage is suspected, there is potential to offer an alternative endovascular approach that obliterates bleeding GVs with or without TIPS creation. ⁸⁶

TIPS Technical Considerations and Postprocedural Care

Since the conceptualization of TIPS over 40 years ago, technical considerations for shunt creation have been described thoroughly and revisited in detail. ^{87 88 89} TIPS creation commonly involves the formation of a tract between the right hepatic vein and right portal vein. This is most commonly achieved by selecting the right hepatic vein and utilizing wedged hepatic venography to identify the right portal vein. A needle is fluoroscopically targeted toward a right portal vein branch. The PSG may be calculated once successful portal venous access has been achieved. Contemporary TIPS are generated by deployment of a stent graft (Viatorr; W.L. Gore, Flagstaff, AZ) for shunt creation. Bare metal stents are rarely used in modern TIPS procedures, given reduced overall patency and association with increased rebleeding rates and decreased survival ^{87 90 91 92 93 94} ; nonetheless, there may be specific clinical scenarios that merit consideration of bare metal stent TIPS creation ( Fig. 3 ). After stent-graft deployment, the shunt tract is dilated incrementally with an 8- to 10-mm balloon to a target PSG ≤ 12 mm Hg. ⁸⁹ Additional splenic and portal venography can be performed to detect varices, which may serve as targets for embolization, most commonly the left gastric (coronary) vein, posterior gastric vein, and short gastric vein. ^{89 95} Moreover, if GVs are present, the operator may choose to obliterate varices at the time of TIPS creation or pursue staged obliteration at a later date. ⁸⁶ Importantly, balloon tamponade devices need to be deflated after shunt creation to ensure variceal decompression and to allow catheter access to varices if embolization or obliteration is to be pursued. Another important consideration when treating acute VH is preprocedural correction of coagulopathy, ⁹⁶ which is often not pursued in the cirrhotic patient prior to salvage therapy.

Fig. 3.

Bare metal stent TIPS in the setting of bacteremia. A 63-year-old man transferred for management of actively bleeding EVs after unsuccessful banding. Patient presented with fever, leukocytosis, and computed tomography findings of colitis. There was clinical concern for bacteremia, though culture results were pending. TIPS was created using bare metal stent ( a ; arrow ) to avoid potential risk of stent-graft colonization, with plan to revise stent-graft TIPS when blood cultures confirmed to be sterile. After TIPS, bleeding stopped, and patient returned 5 days later for successful shunt relining using PTFE-covered stent graft ( b ; arrow ). Infection of TIPS stent grafts—termed endotipsitis—is an infrequent but serious complication associated with high mortality rates. As such, bacteremia or uncontrolled sepsis is considered a relative contraindication for TIPS creation. In such scenarios, bare metal stent may offer lower risk of colonization, given absence of covering graft material.

In the post-TIPS period, the acute VH patient is generally monitored in the intensive care unit to monitor for hemodynamic stability, resolution of hemorrhage, TIPS-related adverse events, and neurological signs of HE. ⁸⁶ Alongside monitoring nasogastric tube output, serial hemoglobin and hematocrit values may be obtained to monitor cessation of VH. A follow-up liver panel and coagulation panel are also obtained the day after the procedure to assess hepatic function. TIPS patency may be monitored with serial Doppler ultrasound at 1, 3, and 6 months postprocedure.

TIPS Creation for Initial Control of VH

TIPS creation is nearly always technically feasible and is a bona fide treatment for initial control of acute VH, ^{75 97} as historical datasets report rates of hemostasis between 89 and 100%, ^{98 99 100 101 102 103 104 105 106 107 108 109} largely in EV hemorrhage. However, historical cohorts have demonstrated variable rebleeding rates, ⁹⁷ which may be inflated due to early widespread utilization of uncovered bare metal stents prior to the advent of the bile impermeable polytetrafluoroethylene (PTFE) stent grafts, which prevent bile-mediated thrombosis and TIPS dysfunction. ¹¹⁰ For instance, a recent retrospective cohort demonstrated significantly lower rebleeding rates (14 vs. 37%, odds ratio [OR] = 0.259; p  < 0.001) and need for shunt revision when comparing PTFE TIPS stent grafts with bare metal stents. ⁹³

Historically, studies have reported variable short-term mortality rates for salvage TIPS, ranging between 19 and 60%. ⁹⁷ Tenuous outcomes continue to be reported for salvage TIPS in hemodynamically unstable patients. Active hemorrhage at the time of TIPS creation has been associated with a 2.9-fold greater odds of death. ^{74 111} A recent retrospective analysis reported approximately 34% mortality at 30 days in patients with hemorrhagic shock undergoing salvage TIPS, in addition to overall unfavorable clinical outcome. ¹¹²

While studies examining TIPS in acute VH are variable due to inclusion criteria and methodology, recent studies demonstrate similar outcomes for rebleeding and overall survival. Gaba et al reported 1-month rebleeding rates less than 10% with rebleeding rates of 22% at 1 year and nearly 30% at 2 years, and survival rates of 80, 69, and 65% at 30 days, 1-year, and 2 years, respectively. ¹¹³ Zhu et al recently reported similar findings, including rebleeding rates of 11, 17, and 20% at 6 weeks, 1 year, and 2 years, respectively. ¹¹⁴ Transplant-free survival was reported to be 88, 82, and 74% at 6 weeks, 1 year, and 2 years, respectively. ¹¹⁴ Longer term survival data remain sparse. Notably, Sanyal et al reported 46% survival at a median follow-up period of 920 days after salvage TIPS, while Azoulay et al reported 1-year survival of 51 and 40% at 3 and 5 years ^{98 105} ; however, these datasets preceded the era of PTFE-covered stent grafts.

Additionally, comparison between emergent TIPS and creation of a surgical portosystemic shunt (SPSS) remains debated. A recent systematic review demonstrated low confidence when assessing RCTs that demonstrated higher rates of all-cause mortality (RR: 0.61, 95% CI: 0.42–0.90), rebleeding (RR: 0.19, 95% CI: 0.08–0.49), reintervention (RR: 0.13, 95% CI: 0.06–0.28), and shunt occlusion (RR: 0.14, 95% CI: 0.040–0.51) at 5 years with TIPS versus SPSS. Given the high numbers of bare metal stents utilized in available datasets, ^{115 116 117 118} prospective data exclusively utilizing PTFE-covered TIPS are necessary.

TIPS Creation for GV Hemorrhage

TIPS for GV hemorrhage requires more data. TIPS has been reported to achieve bleeding control in 90% of bleeding GVs ^{107 119} and, in a RCT, demonstrated significantly lower rebleeding from GV hemorrhage when compared with endoscopic glue obliteration at a median follow-up of 33 months. ¹²⁰ However, recent data suggest that TIPS may not be the optimal modality for addressing GV hemorrhage, which bleed at lower pressures with recent data demonstrating a rebleeding rate of 27 and 15% mortality at 90 days. ¹²¹ Therefore, patients presenting with GV hemorrhage may be considered for concurrent obliteration in addition to TIPS. ⁸⁶

Early TIPS Creation for Acute VH

Early nonsalvage TIPS creation (usually within 72 hours) is an emerging indication in VH with promising results. Early TIPS supplements initiated pharmacologic and endoscopic interventions, and is theorized to mitigate background elevated portal pressure and prevent rebleeding. ¹²² Initially, a RCT with 116 consecutive patients utilized uncovered stents in early VH (24 hours), and demonstrated that patients with HPVG ≥ 20 mm Hg had significantly less treatment failure (12 vs. 50%) and transfusion requirements when compared with those not undergoing TIPS. ¹²³ This study also reported significantly lower in-hospital and 11-year mortality (11 and 31%, respectively). ¹²³ Subsequently, a RCT of patients with Child-Pugh B/C VH showed improved rates of bleeding control (97 vs. 50%), 6-week survival (97 vs. 67%), and 1-year actuarial survival (86 vs. 61%) with a PTFE-covered TIPS versus pharmacologic and endoscopic therapy. ¹²⁴ A subsequent retrospective review comparing TIPS versus combination pharmacotherapy and endoscopy in the same center demonstrated similar findings, as the TIPS group experienced significantly lower rates of failure to control bleeding or rebleeding (93 vs. 50%) and improved 1-year actuarial survival (86 vs. 70%). ¹²⁵ Another prospective study aimed at external validation of initial results, and demonstrated significantly lower 1-year probability for rebleeding with early TIPS creation in a similar patient cohort (97 vs. 51%; p  < 0.001), but reported no difference in actuarial survival. ¹²⁶

Recent data are also available. A multicenter observational study comparing early TIPS and combination pharmacotherapy and endoscopy in 631 patients demonstrated significantly improved survival at 1 year when comparing early TIPS to combination pharmacotherapy and endoscopy in patients with Child-Pugh C disease along with reduced treatment failure (which incorporated rebleeding). ⁶⁰ Additionally, a recent RCT examining early covered TIPS with pharmacologic therapy and endoscopy examined transplant-free survival as an endpoint and demonstrated significantly improved transplant-free survival at 6 weeks and 1 year in patients who received early TIPS. ¹²⁷

Several meta-analyses have also been performed. An analysis from 2015 evaluated randomized and nonrandomized data in early TIPS creation, reporting fewer deaths (OR: 0.38, 95% CI = 0.17–0.83; p  = 0.02) and lower rates of rebleeding within 1 year in patients with Child-Pugh B (OR: 0.08, 95% CI: 0.04–0.17; p  < 0.001) and Child-Pugh C (OR = 0.05, 95% CI = 0.02–0.15; p  < 0.001) disease. ¹²⁸ A subsequent meta-analysis echoed these findings, reporting significantly decreased inpatient mortality (RR = 0.87) and rebleeding (RR = 0.56). ¹²⁹ Notably, early TIPS did not demonstrate increased HE in these studies, which is an important consideration for patients, families, consulting physicians, and operators.

Additionally, a 10-year retrospective inpatient dataset of 142,539 patients with decompensated cirrhosis and EV hemorrhage demonstrated significantly decreased rebleeding (RR = 0.56) and inpatient mortality (RR = 0.87) with early TIPS when compared with no TIPS. ¹³⁰ Significantly less inpatient rebleeding, death, and cost were observed as well. ¹³⁰ Additional analysis compared early TIPS versus salvage TIPS, yielding significantly less inpatient rebleeding, death, length of stay, and cost for the early TIPS group. ¹³⁰ These latter results may reflect the dire clinical circumstances encountered when salvage TIPS is considered.

TIPS Creation with Adjunctive Embolotherapy

Adjunctive embolotherapy of varices is a protective measure used alongside TIPS creation, and is theorized to not only result in variceal occlusion, but may also serve to maintain TIPS patency by eliminating competing variceal shunts which could contribute to TIPS dysfunction and promote subsequent GEV formation and VH. ^{97 131 132} The embolization procedure may be staged prior to or following TIPS creation. ⁸⁶ Post-TIPS variceal embolization allows the operator to view optimized filling of the newly created TIPS and its impact on variceal filling. Alternatively, preprocedural embolotherapy has the benefit of improved variceal filling to guide targeting. Moreover, embolotherapy performed before TIPS creation avoids an avenue for embolic agent migration and nontarget embolization.

Although controversial, adjunctive embolotherapy has been utilized in up to 48% of patients ^{97 131} and has largely favorable results considering prospective and retrospective data. ⁸⁶ Prospective studies have shown significant reductions in rebleeding rates with adjunctive embolotherapy when compared with TIPS alone. ^{132 133 134} Retrospective data are mixed, with a majority of studies with significant reductions in rates of rebleeding when adjunctive embolotherapy is utilized. ^{135 136} Other retrospective data demonstrate lower, although nonsignificant, rates of rebleeding between combined treatment and TIPS alone. ^{88 137 138} Moreover, a meta-analysis comparing combined TIPS and adjunctive embolotherapy versus TIPS alone reported that combined therapy significantly reduces rebleeding and improves TIPS patency at 6 months. ¹³⁹ A recent retrospective study exploring adjunctive embolotherapy reported nearly 6% rebleeding rate at a median follow-up time of 26 months with approximately 1 and 3% rebleeding rates at 1 and 2 years. ¹⁴⁰

Conclusion

Patients with decompensated liver cirrhosis and acute VH require multifaceted, multidisciplinary treatment. Emergent TIPS creation successfully ameliorates acute VH when standard medical and endoscopic management are unsuccessful, and recent data demonstrate a promising role for early, nonsalvage TIPS creation in acute VH. Patient selection and risk stratification for this potentially life-saving procedure remains challenging and continues to be determined on a case-by-case basis. Decision making can be aided by objective measures such as MELD score along with anatomic considerations such as the presence of bleeding GVs which could benefit from obliteration or combined TIPS approaches.