Abstract
BACKGROUND: Hepatic encephalopathy (HE), especially overt HE (OHE), is the most common transjugular intrahepatic portosystemic shunt (TIPS) complication, affecting quality of life. Existing noninvasive models have suboptimal predictive performance. AIM: To develop a novel model integrating preoperative and intraoperative indicators to predict post-TIPS OHE. METHODS: We included 455 TIPS patients from three centers (Army Medical Center, Xinqiao Hospital, and West China Hospital) between March 2018 and September 2024, randomly allocated into 7:3 development and validation cohorts. LASSO and multivariate logistic regression identified independent risk factors, with model discrimination/calibration evaluated against traditional models (Child-Pugh, model for end-stage liver disease [MELD], albumin-bilirubin [ALBI], and Freiburg index of post-TIPS survival [FIPS]). RESULTS: A total of 455 patients were included. During follow-up, 91 patients (28.4%) in the development cohort and 41 (30.4%) patients in the validation cohort developed OHE. The analysis identified the following independent risk factors: peripheral blood monocyte count (odds ratio [OR] = 1.40, 95% confidence interval [CI]: 1.22-1.63, P < 0.001), thrombin time (TT; OR = 1.35, 95% CI: 1.15-1.59, P < 0.001), age (OR = 1.08, 95% CI: 1.05-1.12, P < 0.001), history of diabetes (OR = 7.64, 95% CI: 3.50-17.55, P < 0.001), ascites grade [mild-moderate ascites (OR = 2.35, 95% CI: 1.19-4.77, P = 0.02), severe ascites (OR = 3.76, 95% CI: 1.15-12.64, P = 0.03)]. Stent underdilated (8-mm stent dilated to 6-mm) was a strong independent protective factor (OR = 0.12, 95% CI: 0.05-0.26, P < 0.001). The six-factor prediction model achieved area under the curve values of 0.864 (development cohort) and 0.812 (validation cohort), significantly outperforming Child-Pugh (0.617/0.656), MELD (0.551/0.569), ALBI (0.607/0.662), and FIPS (0.667/0.608) scores (all P < 0.05). CONCLUSION: The prediction model incorporating stent dilatation status, peripheral blood monocyte count, TT, age, ascites severity, and history of diabetes demonstrates high predictive accuracy for assessing OHE risk after TIPS. Notably, the model was derived in a mainly hepatitis B virus-cirrhosis Asian population, which may limit generalizability to alcoholic/nonalcoholic steatohepatitis-dominant western populations.