On the basis of clinical and laboratory data, researchers sought to create pediatric-specific models that predicted liver stiffness and hepatic steatosis in non-alcoholic fatty liver disease (NAFLD). The study included children with NAFLD who had undergone magnetic resonance imaging with proton density fat fraction (MRI-PDFF) for steatosis quantification and/or magnetic resonance elastography (MRE) for liver stiffness evaluation. To construct a prediction model for fat percentage and stiffness, they collected data from patients scanned between April 2009 and July 2018. They used data from a second cohort, photographed between 2018 and 2019, to confirm the models’ performance.

The first cohort (n=344) was primarily made up of non-Hispanic (80%), male (67%) teenagers. There were 343 children with MRE data and 130 with PDFF data. Ethnicity, insulin levels, platelet count, and aspartate aminotransferase all independently predicted liver stiffness in multivariable regression, and these factors were utilized to build the prediction model. Similarly, sex, ethnicity, alanine aminotransferase, and triglyceride levels all predicted liver PDFF separately and were utilized in the PDFF model. The AUC of the best cutoff for the model that predicted stiffness more than 2.71 kPa was 0.70, whereas the AUC for the model that predicted PDFF greater than 5% was 0.78. Similar features were found in the validation group (n=110). The model’s correlation coefficient with measured liver stiffness and PDFF was 0.30 and 0.26, respectively. 

Pediatric-specific models predicted precise liver stiffness and steatosis poorly; nevertheless, in the absence of magnetic resonance imaging, they could be used to predict the presence of severe steatosis (>5%) and/or significant stiffness (>2.71). As a result, imaging remains a crucial supplement to laboratory tests in evaluating disease severity. 

Reference:journals.lww.com/jpgn/Abstract/2022/04000/Non_Invasive_Approaches_to_Estimate_Liver.14.aspx