Treatment Outcomes and Predictors After Endovascular Intervention for Portal Hypertension-Related Gastrointestinal Bleeding in Cirrhosis

Portal hypertension-related gastrointestinal bleeding is a major clinical problem in patients with cirrhosis and may require endovascular treatment when bleeding is severe, recurrent, refractory to conventional therapy, or associated with complex portosystemic vascular anatomy. Endovascular management includes decompressive procedures such as transjugular intrahepatic portosystemic shunt (TIPS) and obliteration techniques such as coil-assisted or plug-assisted retrograde transvenous obliteration (CARTO/PARTO) and antegrade transvenous obliteration (ATO). The choice of procedure is influenced by the bleeding presentation, variceal anatomy, portosystemic collateral pathways, portal vein patency, hepatic function, and the clinical condition of the patient.

The clinical course following these interventions may vary substantially according to baseline liver function, severity of the acute bleeding episode, vascular anatomy, and the type of intervention performed. Important outcomes include achievement of initial bleeding control, recurrent bleeding, procedure-related complications, hepatic encephalopathy, worsening ascites, need for reintervention, and mortality. However, the relationship between pre-procedural characteristics and these outcomes has not been systematically evaluated in a broad Vietnamese cohort undergoing different endovascular techniques.

The TOPIC study is a hybrid observational cohort conducted at the Department of Gastroenterology and the Endovascular Intervention Unit of the University Medical Center Ho Chi Minh City. The study includes two components: a retrospective cohort of eligible patients treated from January 2021 through September 2026 and a prospective cohort of patients consecutively enrolled from October 2026 through March 2028. The study population consists of adults with confirmed cirrhosis who present with acute gastrointestinal bleeding attributable to portal hypertension and undergo at least one endovascular intervention, including TIPS, PARTO, CARTO, ATO, or combined procedures. Pregnant or lactating women, patients without confirmed cirrhosis, patients with bleeding unrelated to portal hypertension, and patients with insufficient medical records for assessment of the primary outcome will be excluded.

For retrospectively identified participants, clinical information will be obtained from electronic medical records, imaging and PACS systems, endoscopy records, and digital subtraction angiography databases. For prospectively enrolled participants, information will be collected using standardized case report forms during hospitalization, with follow-up at six weeks after discharge through clinical assessment or structured telephone contact. Baseline information will include demographic characteristics, cirrhosis etiology, history of portal hypertensive bleeding or hepatic encephalopathy, ascites, characteristics of the bleeding episode, endoscopic findings, cross-sectional imaging findings, portal and systemic venous anatomy, and the presence of portal vein thrombosis.

Laboratory and disease-severity variables will be obtained using the most recent measurements available before the endovascular procedure. These include hemoglobin, platelet count, INR, albumin, total bilirubin, creatinine, serum sodium, aminotransferases, inflammatory markers, and, when available, ammonia and NT-proBNP. Several established liver disease severity scores will be calculated, including Child-Pugh, MELD, MELD-Na, MELD 3.0, and CLIF-C AD. The study will specifically examine whether these pre-procedural measures provide clinically useful information for identifying patients at higher risk of adverse outcomes.

Procedural characteristics will include the endovascular technique, urgency of intervention, vascular access route, embolic materials, TIPS stent diameter when applicable, portal venous pressure measurements, and portosystemic pressure gradient measurements when available. Technical success will be assessed according to the procedure performed. For TIPS, this requires establishment of a functioning shunt, whereas for embolization procedures it requires complete or near-complete occlusion of the targeted varices or shunt on final angiography.

Clinical outcomes will be assessed after the intervention. Initial bleeding control is defined by absence of ongoing bleeding, hemodynamic stabilization, and no requirement for repeat intervention or endoscopy during the first 72 hours. Rebleeding will be identified on the basis of recurrent clinical bleeding, acute anemia, endoscopic or laboratory evidence, transfusion requirements, or repeat intervention. Other outcomes include in-hospital mortality, clinical success, post-intervention hepatic encephalopathy, worsening ascites, worsening esophageal varices, post-procedural complications, reintervention, length of hospital stay, and six-week post-discharge mortality in the prospective cohort.

The principal prognostic analysis will evaluate associations between pre-procedural characteristics and adverse clinical outcomes. Candidate predictors will include demographic characteristics, clinical presentation, laboratory parameters, liver disease severity scores, vascular anatomy, and characteristics of the planned intervention. Logistic regression will be used for binary outcomes, while Cox proportional hazards models will be applied to time-to-event outcomes. Kaplan-Meier methods will be used to describe event-free survival and mortality over time. The discriminative performance of Child-Pugh, MELD, MELD-Na, MELD 3.0, and CLIF-C AD will also be compared using receiver operating characteristic analysis and the area under the curve.

The study is designed to provide a structured clinical dataset covering different endovascular approaches within a single tertiary-care center. By examining both procedural results and patient-level factors present before intervention, the study will characterize heterogeneity in clinical outcomes and explore variables that may assist future risk stratification and selection of endovascular treatment strategies in patients with cirrhosis and portal hypertension-related gastrointestinal bleeding.