Liver fibrosis scores in the general population: Better risk indices are needed!

Potential conflict of interest: Nothing to report. TO THE EDITOR: Unalp‐Arida and Ruhl are to be congratulated on their analysis of liver fibrosis scores in predicting liver mortality in the unselected general population.1 With the increasing population burden of liver disease, especially nonalcoholic fatty liver disease (NAFLD), it is a great unmet need to be able to detect individuals in the general population that are at high risk of developing clinical, complicated liver disease, in order to target a more intense diagnostic liver workup, possible specific treatment, and closer follow‐up to these individuals. Liver enzymes or simple steatosis alone are insufficient given that only a small fraction of those with elevated transaminases or steatosis will ever develop clinical liver disease. With the degree of liver fibrosis being the strongest prognostic factor, estimates of liver fibrosis based on readily available laboratory measurements are attractive. Unalp‐Arida and Ruhl1 found increased liver mortality among individuals with high fibrosis risk as estimated by aspartate aminotransferase to platelet ratio index (APRI), NAFLD fibrosis score, Fibrosis‐4 (FIB‐4), and Forns score. However, among subjects deemed to have such high fibrosis risk, the proportion of deaths attributable to liver disease was small: 4% for NAFLD fibrosis score, 7% for FIB‐4, and 19% for APRI (intermediate and high risk combined because of small numbers!). Moreover, the proportion of liver deaths captured by a high fibrosis risk was also quite unsatisfactory: 34% for APRI (intermediate to high), 24% for FIB‐4, and 25% for NAFLD fibrosis score. Taking into account the population prevalence of high fibrosis risk and based on the numbers given in the article, the age‐adjusted population attributable fraction for the various scores is estimated to range from 30% to 40%. This is the proportion of liver deaths in the general population that is attributed to a high fibrosis risk score, as measured in the study. In other words, 60%‐70% of liver deaths are not captured by a high liver fibrosis score. Therefore, none of the scores tested showed sufficient utility to be useful in the unselected general population. Although serial measurements would likely improve the utility of these scores because the study spanned 23 years, an issue also acknowledged by the researchers, there seems to be a need for better risk indices. Such risk indices should ideally incorporate key environmental and genetic drivers of liver disease.2