SIR 2024
Portal Hypertension
Malinda Gong, BS
Medical Student
University of Virginia
Financial relationships: Full list of relationships is listed on the CME information page.
James Patrie, MS
Senior Biostatistician
University of Virginia Health System
Disclosure information not submitted.
Luke R. Wilkins, MD, FSIR
Associate Professor
University of Virginia
Financial relationships: Full list of relationships is listed on the CME information page.
John F. Angle, MD
Professor
University of Virginia
Financial relationships: Full list of relationships is listed on the CME information page.
Evaluate the validity and accuracy of hepatic risk stratification systems in predicting mortality and survival after TIPS.
Materials and Methods:
Single-center retrospective review from January 2017 to July 2021 of all TIPS done in adults. Demographics, indication, liver disease etiology, and pertinent technical and laboratory values were collected. Five mortality risk stratification predictors for hepatic dysfunction were analyzed: three continuous predictors [model for end-stage liver disease (MELD), MELD-Na, and MELD 3.0] and two categorical predictors [Child-Pugh (CP) and albumin-bilirubin (ALBI)]. Univariate and multivariate logistic regression models were used to generate odds ratios (OR) with 95% confidence levels (CL), and a Fisher’s Exact Test performed to evaluate for predictability of mortality.
Results:
Out of 183 patients, 70 were female. Mean age was 57.9±10.6 years. TIPS was classified as elective [127 (69.4%)], urgent (inpatient) [50 (27.3%)], or emergent [6 (3.3%)]. Most common indication was ascites (51.4%) or esophageal varices (24.6%). Most common liver disease etiology was NASH (31.1%) or EtOH (25.7%).
Pre-TIPS risk scores were (mean): MELD (12.8 ± 3.8); MELD-Na (14.8 ± 5.4); MELD 3.0 (15.1 ± 5.2); CP: class A – 7.1%, B - 73.2%, and C - 19.7%; ALBI: grade 1 - 4.5%, 2 - 68.8%, 3 - 26.7%.
Technical success was 99.5% (182/183). Median overall survival was 31.8 months (95% CL, 28.3-36.4). Inpatient mortality was 6.6% (N=12), and 3-month mortality was 15.3% (N=27).
For inpatient mortality, a one-unit score increase demonstrated: MELD OR 1.22 (CL: 1.05-1.43, p = 0.01), MELD-Na OR 1.11 (CL: 1.00-1.24, p = 0.05), and MELD 3.0 OR 1.14 (CL: 1.01- 1.29, p = 0.03). For 3-month mortality, a one-unit score increase demonstrated: MELD OR 1.10 (CL: 0.98-1.22, p = 0.10), MELD-Na OR 1.08 (CL: 1.00-1.17, p = 0.05), and MELD 3.0 OR 1.11 (CL: 1.01- 1.21, p = 0.02).
Inpatient mortality was significantly associated with ALBI grade 3 versus 2 (p = 0.004) but not with CP Class C versus B (p = 0.70). At 3-months, mortality association of ALBI Grade 3 (versus 1/2) showed OR 4.53 (CL: 1.84-11.11, p = 0.001) and CP Class C (versus A or B) demonstrated OR 1.52 (CL: 0.57-4.05, p = 0.40). On multivariate analysis, ALBI Grade 3 (versus 1 or 2) showed persistent significance with OR 6.07 (CL: 1.85-19.94, p = 0.003).
Conclusion: MELD, MELD-Na, MELD 3.0 and ALBI are statistically significant predictors of inpatient mortality after TIPS. MELD-Na, MELD 3.0, and ALBI were superior to MELD and CP in predicting 3-month mortality. ALBI grade has persistent statistically significant predictive value on multivariate analyses.