Development and validation of a noninvasive clinical scoring system to predict significant fibrosis in patients with nonalcoholic fatty liver disease

Abstract Background The aims of this study were to identify risk factors for significant fibrosis (SF) by assessing physical and laboratory parameters and develop and validate a clinical score and nomogram for the prediction of SF in patients with nonalcoholic fatty liver disease (NAFLD). Methods This retrospective study included 225 patients with histologically confirmed NAFLD who were divided into two cohorts using 10-fold cross validation for model training and validation. The clinical score and nomogram were used to predict the NAFLD outcome. Results The model for predicting SF (stage ≥ 2) including the free T4/free T3 ratio, low-density lipoprotein cholesterol, homeostatic model assessment for insulin resistance (HOMA-IR), percentage of appendicular skeletal muscle mass and aspartate aminotransferase (AST) level in the training and validation cohorts yielded an area under the receiver operator characteristic curve (AUROC) of 0.79 and 0.78, respectively. The AUROC of the combined clinical score for the prediction of SF was 0.82 (95% CI, 0.75-0.89) at a cutoff value of 3 points, with a sensitivity (SE) of 77.19%, specificity (SP) of 82.88%, positive predictive value (PPV) of 63.77%, and negative predictive value (NPV) of 90.30%. The nomogram had good performance in quantitatively predicting the risk probability of SF. Conclusion Our study showed that a noninvasive clinical scoring system using easily available physical and laboratory variables can identify patients with NAFLD with or without SF with a high degree of accuracy. Application of this system may decrease the need for staging liver biopsy specimens and allow early identification and intervention in these high-risk patients.