Modified FIPS with nutrition and muscle metrics predicts severe encephalopathy after TIPS: A multicenter study

Background Severe hepatic encephalopathy (SHE) following transjugular intrahepatic portosystemic shunt (TIPS) worsens the prognosis of cirrhotic patients. Despite established associations with nutritional and muscle-related factors, no integrated predictive model has yet been developed. This study developed and validated a model combining the Freiburg post-TIPS survival (FIPS) score with nutrition and sarcopenia-related indicators. Methods This retrospective study included 590 cirrhotic patients who underwent TIPS at four Chinese hospitals between April 2015 and March 2025. Patients from Center I were divided into training and internal validation cohorts at a ratio of 7:3, while those from the remaining centers formed the external validation cohort. Independent risk factors identified via Cox regression were incorporated into a base model using FIPS and preoperative CT-derived nutrition-sarcopenia indices. Model performance was evaluated in terms of discrimination (C-index and AUC), calibration, and clinical utility. Results Independent predictors of post-TIPS SHE included FIPS ( P  = 0.001), myosteatosis ( P  = 0.019), and sarcopenia ( P  < 0.001). The modified model demonstrated AUCs of 0.766, 0.745, and 0.751 at 3, 6, and 12 months in the training cohort; 0.733, 0.718, and 0.793 in the internal validation cohort; and 0.811, 0.797, and 0.854 in the external validation cohort, respectively, outperforming the original FIPS score. Predictive accuracy improved by 4.4% ( P  = 0.001), with calibration and decision curve analyses confirming its clinical utility. Conclusion FIPS, sarcopenia, and myosteatosis were identified as independent predictors of post-TIPS SHE. The modified predictive model demonstrated superior performance and clinical utility than that obtained using the conventional FIPS score, providing a reliable tool for risk stratification and management in cirrhotic patients undergoing TIPS.