A model for predicting poor survival in patients with cirrhosis undergoing portosystemic shunt embolization

Background and Aim: Portosystemic shunt embolization (PSSE) is a promising treatment for hepatic encephalopathy (HEP) and gastric varix (GV) in cirrhotic patients with a spontaneous portosystemic shunt. However, PSSE may worsen portal hypertension causing hepatorenal syndrome, liver failure, and mortality. This study aimed to develop and validate a prognostic model that helps identify patients with a risk of poor short-term survival after PSSE. Methods: We included 188 patients who underwent PSSE for recurrent HEP or GV at a tertiary center in Korea. To develop a prediction model for 6-month survival after PSSE, Cox proportional-hazard model was used. The developed model was validated in a separate cohort of 184 patients from two other tertiary centers. Results: In multivariable analysis, the 1-year overall survival after PSSE was significantly associated with baseline levels of serum albumin, total bilirubin, and international normalized ratio (INR). We therefore developed the albumin-bilirubin-INR (ABI) score by assigning 1 point each for albumin < 3.0 g/dL, total bilirubin ≥ 1.5 mg/dL, and INR ≥ 1.5. Time-dependent areas under the curve of the ABI score for predicting 3-month and 6-month survival were 0.85 and 0.85 in the development cohort and 0.83 and 0.78 in the validation cohort, indicating good discrimination performance. The ABI score showed a better discrimination and calibration performance than the model for end-stage liver disease and the Child–Pugh scores, especially in high-risk patients. Conclusions: The ABI score is a simple prognostic model that helps decide whether to proceed with PSSE for the prevention of HEP or GV bleeding in patients with spontaneous portosystemic shunt.