Modified model incorporating sarcopenia and myosteatosis for predicting severe hepatic encephalopathy after transjugular intrahepatic portosystemic shunt: Multicenter study

Figure 1 A visually intuitive nomogram model was constructed based on three significant independent variables: Freiburg index of post-transjugular intrahepatic portosystemic shunt survival score, sarcopenia, and myosteatosis. Each variable was assigned a corresponding score. The total score was calculated by summing these values to predict the risk of post-transjugular intrahepatic portosystemic shunt (TIPS) severe hepatic encephalopathy in patients with cirrhosis. The nomogram is based on the following log-hazard equation: Log-hazard = (β1 × Freiburg index of post-TIPS survival score) + (β2 × sarcopenia) + (β3 × myosteatosis), where β1 = 0.598, β2 = 0.965, and β3 = 0.659. In addition, the baseline hazard function h0 (t) = 0.177, where t = 12. For example, a patient without sarcopenia and myosteatosis, with a Freiburg index of post-TIPS survival score of -2.5, would have a total score of 53 + 53 + 39 = 145. This corresponds to a 2.94% risk of developing severe hepatic encephalopathy at 3 months post-TIPS, a 3.49% risk at 6 months, and a 4.29% risk at 1 year. FIPS: Freiburg index of post-transjugular intrahepatic portosystemic shunt survival.

Figure 2 Time-dependent concordance index curves of the nomogram and Freiburg index of post-transjugular intrahepatic portosystemic shunt survival score for evaluating discrimination for post-transjugular intrahepatic portosystemic shunt severe hepatic encephalopathy over time. A: Training cohort; B: Internal validation cohort; C: External validation cohort. The modified model demonstrated superior performance compared to the Freiburg index of post-transjugular intrahepatic portosystemic shunt survival scoring system in predicting post-transjugular intrahepatic portosystemic shunt severe hepatic encephalopathy across all study cohorts. FIPS: Freiburg index of post-transjugular intrahepatic portosystemic shunt survival.

Figure 3 Receiver operating characteristic curve of the nomogram and Freiburg index of post-transjugular intrahepatic portosystemic shunt survival score predicting post-transjugular intrahepatic portosystemic shunt severe hepatic encephalopathy at 3-month, 6-month, and 1-year. A-C: Training cohort; D-F: Internal validation cohort; G-I: External validation cohort. The modified model exhibited better performance than the Freiburg index of post-transjugular intrahepatic portosystemic shunt survival scoring system in predicting post-transjugular intrahepatic portosystemic shunt severe hepatic encephalopathy at all time points across all study cohorts. AUC: Area under the receiver operating characteristic curve; FIPS: Freiburg index of post-transjugular intrahepatic portosystemic shunt survival.

Figure 4 Calibration curves of the nomogram showing the 3-month, 6-month, and 1-year prediction accuracy for post-transjugular intrahepatic portosystemic shunt severe hepatic encephalopathy. A: Training cohort; B: Internal validation cohort; C: External validation cohort. The nomogram predictions closely aligned with the observed outcomes in all cohorts, demonstrating excellent calibration.

Figure 5 Decision curve analysis of the nomogram and Freiburg index of post-transjugular intrahepatic portosystemic shunt survival score evaluating 3-month, 6-month and 1-year net benefit. A-C: Training cohort; D-F: Internal validation cohort; G-I: External validation cohort. Across all cohorts, the nomogram consistently achieved higher net benefits than the Freiburg index of post-transjugular intrahepatic portosystemic shunt survival scoring system over a wide range of clinical decision thresholds. FIPS: Freiburg index of post-transjugular intrahepatic portosystemic shunt survival.

Figure 6 Comparison of the cumulative incidence rates of post-transjugular intrahepatic portosystemic shunt severe hepatic encephalopathy between the higher-risk and lower-risk groups stratified by the modified predictive model. A: Training cohort; B: Internal validation cohort; C: External validation cohort. The incidence rates of severe hepatic encephalopathy in the training cohort were 41/118 (34.7%) for the higher-risk group and 16/177 (9.0%) for the lower-risk group. In the internal validation cohort, the incidence rates were 13/37 (35.1%) for the higher-risk group and 8/87 (9.2%) for the lower-risk group. In the external validation cohort, the incidence rates were 19/45 (42.2%) for the higher-risk group and 11/126 (8.7%) for the lower-risk group.