Establishment of a model for evaluating the severity of nonalcoholic fatty liver disease based on transient elastography parameters

• Main Author: YAO Mingjie
• Format: Article
• Language: zho
• Published: Editorial Department of Journal of Clinical Hepatology 2021-07-01
• Series: Linchuang Gandanbing Zazhi
• Online Access: http://www.lcgdbzz.org/cn/article/doi/10.3969/j.issn.1001-5256.2021.07.027

ObjectiveTo establish a model for evaluating the severity of nonalcoholic fatty liver disease (NAFLD) based on a combined analysis of transient elastography parameters. MethodsA retrospective analysis was performed for the clinical information and liver elasticity parameters of 184 NAFLD patients who attended 7 hospitals in China from July 2014 to July 2017 and underwent liver biopsy for pathological diagnosis. Liver elasticity parameters were named as P1-P18 according to the amplitude, frequency, dispersion, and attenuation of ultrasound signals. The Spearman rank correlation test and the linear regression analysis were used to establish the models for evaluating steatosis degree, inflammatory activity, and fibrosis degree, and the receiver operating characteristic (ROC) curve was used to evaluate the value of each diagnostic model. ResultsThe ultrasound parameters based on transient elastography were well correlated with steatosis degree, inflammatory activity, and liver fibrosis degree in NAFLD patients. The multivariate analysis showed that P1 and P12 were independent indicators for predicting steatosis degree; P1, P3, P6, and P11 were independent indicators for predicting inflammatory activity; P2 and P3 were significantly correlated with liver fibrosis degree. The model Y=0.013×P1+0.055×P12-0.318 5 for predicting steatosis degree established based on linear regression had an area under the ROC curve (AUC) of 0.895 (95% confidence interval ［CI］: 0.842-0.936) and 0.939 (95%CI: 0.894-0.969), respectively, in the diagnosis of mild steatosis and moderate-to-severe steatosis, with a sensitivity of 82.35% and 86.26%, respectively, and a specificity of 89.23% and 93.27%, respectively (P＜0.001). The model Y=0.008×P1+0.030×P3+0.029×P6-1.875×10-4×P11+0.416 for predicting inflammatory activity had an AUC of 0.828 (95%CI: 0.793-0865) and 0.874 (95%CI: 0.817-0.918), respectively, in the diagnosis of early-stage inflammation and significant inflammation, with a sensitivity of 70.6% and 73.96%, respectively, and a specificity of 85.7% and 93.33%, respectively (P＜0.001). The model Y=-0.003×P1 + 0.601×(lnP2)+0.285×(lnP3)+0.036×P15 + 0.078 for predicting liver fibrosis degree had an AUC of 0.805 (95%CI: 0.740-0.869), 0.767 (95%CI: 0.699-0.827), and 0.803 (95%CI: 0.701-0.906), respectively, in the diagnosis of significant fibrosis, severe fibrosis, and early liver cirrhosis, with a sensitivity of 72.27%, 77.63%, and 90.00%, respectively, and a specificity of 79.69%, 66.42%, and 66.47%, respectively (P＜0.001). ConclusionDifferent parameters of transient elastography can effectively reflect steatosis degree, inflammatory activity, and fibrosis degree in NAFLD patients, and a combined model can improve the accuracy of disease severity prediction in NAFLD patients.