An alternative model for prediction of transplant-free survival in patients with implantation of transjugular intrahepatic portosystemic shunt (TIPS): The Freiburg risk score

Introduction:

Implantation of transjugular intrahepatic portosystemic shunt (TIPS) is an effective and safe treatment for complications of portal hypertension in patients with liver cirrhosis. Liver function is an important prognostic factor in these patients. Recently, the albumin-bilirubin (ALBI) grade has been introduced for assessment of liver function in patients with hepatocellular carcinoma and it is a useful tool for objective assessment of liver function in these patients. However, several studies showed that the ALBI score may not be a useful prediction model for patients with decompensated liver cirrhosis and TIPS implantation. Therefore, the aim of our study was to develop a modified ALBI score in order to increase its prognostic accuracy in these patients.

Patients and methods:

895 patients with de novo TIPS implantation (Freiburg: n = 397, Münster: n = 104; Ludwigsburg: n = 134, Heidelberg: n = 260) were included in the analyses. The study cohort was divided in a training set (n = 314), an internal validation set (n = 187) and an external validation set (n = 394). Medical records, laboratory parameters and transplant-free survival (TFS) were assessed. Cox regression model was performed to create an alternative prediction model which includes the ALBI score.

Results:

For assessment of prognostic factors a Cox regression model including the ALBI score, creatinine, logINR and portosystemic gradient was performed and creatinine (HR1.36 1.21 – 1.53, p < 0.001] and the ALBI score (HR 1.23 [1.02 – 1.47]) emerged as significant prognostic factors. By using the corresponding regression coefficient of the significant variables, we created a new model for prediction of transplant-free survival after TIPS implantation which was named the Freiburg Risk score (FRS) as a modification of the ALBI score. The FRS can be calculated according to the following formula: FRS = 0.31*creatinine + 0.20* ALBI score. By using the 85th percentile of the continuos score a high risk group with a significantly reduced prognosis was identified. Indeed, this group showed a TFS of 7 [4.4 – 9-9] months in the training and 2 [0 – 4.6] months in the internal and 2 [0.6 – 3.4] months in the external validation set. In comparison to the Child, MELD and especially the ALBI score the FRS showed good prognostic discrimination assessed by the c-index (FRS: 0.726, Child:0.667, MELD: 0.694, ALBI: 0.573).

Conclusion:

The Freiburg risk score (FRS) as a modification of the ALBI score may be an alternative prediction model for TFS after TIPS implantation and is able to identify high risk patients with a reduced prognosis after TIPS implantation.