Integrated analyses and a novel nomogram for the prediction of significant fibrosis in patients

This study aimed to explore the key genes involved in the pathophysiological process of liver fibrosis and develop a novel predictive model for noninvasive assessment of significant liver fibrosis patients.

Differentially expressed genes (DEGs) were identified using the Limma package. The hub genes were explored using the CytoHubba plugin app and validated in GEO datasets and cell models. Furthermore, serum LTBP2 was measured in liver fibrosis (LF) patients with biopsy-proven by ELISA. All patients' clinical characteristics and laboratory results were analyzed. Finally, multivariate logistic regression analysis was used to construct the model for visualization by nomogram. Area under the receiver operating characteristic curve (AUROC) analysis, calibration curves, and decision curve analysis (DCA) certify the accuracy of the nomogram.

RNA sequencing was performed on the liver tissues of 66 biopsy-proven HBV-LF patients. After multiple analyses and in vitro simulation of HSC activation, LTBP2 was found to be the most associated with HSC activation regardless of the causes. Serum LTBP2 expression was measured in 151 patients with biopsy, and LTBP2 was found to increase in parallel with the fibrosis stage. Multivariate logistic regression analysis showed that LTBP2, PLT and AST levels were demonstrated as the independent prediction factors. A nomogram that included the three factors was tabled to evaluate the probability of significant fibrosis occurrence. The AUROC of the nomogram model was 0.8690 in significant fibrosis diagnosis.

LTBP2 may be a new biomarker for liver fibrosis patients. The nomogram showed better diagnostic performance in patients.