Predicting Post-Hepatectomy Liver Failure in HCC Patients: A Review of Liver Function Assessment Based on Laboratory...

The assessment of liver function is crucial in predicting the risk of post-hepatectomy liver failure (PHLF) in patients undergoing liver resection, especially in cases of hepatocellular carcinoma (HCC) which is often associated with cirrhosis. There are currently no standardized criteria for predicting the risk of PHLF. Blood tests are often the first- and least invasive expensive method for assessing hepatic function. The Child-Pugh score (CP score) and the Model for End Stage Liver Disease (MELD) score are widely used tools for predicting PHLF, but they have some limitations. The CP score does not consider renal function, and the evaluation of ascites and encephalopathy is subjective. The MELD score can accurately predict outcomes in cirrhotic patients, but its predictive capabilities diminish in non-cirrhotic patients. The albumin-bilirubin score (ALBI) is based on serum bilirubin and albumin levels and allows the most accurate prediction of PHLF for HCC patients. However, this score does not consider liver cirrhosis or portal hypertension. To overcome this limitation, researchers suggest combining the ALBI score with platelet count, a surrogate marker of portal hypertension, into the platelet-albumin-bilirubin (PALBI) grade. Non-invasive markers of fibrosis, such as FIB-4 and APRI, are also available for predicting PHLF but they focus only on cirrhosis related aspects and are potentially incomplete in assessing the global liver function. To improve the predictive power of the PHLF of these models, it has been proposed to combine them into a new score, such as the ALBI-APRI score. In conclusion, blood test scores may be combined to achieve a better predictive value of PHLF. However, even if combined, they may not be sufficient to evaluate liver function and to predict PHLF; thus, the inclusion of dynamic and imaging tests such as liver volumetry and ICG r15 may be helpful to potentially improve the predictive capacity of these models.