Distinguishing between hepatocellular adenoma and well-differentiated hepatocellular carcinoma using MRI and clinical feature-based nomogram model.

[AIM] To assess MRI and clinical features for the differentiation of hepatocellular adenoma (HCA) and well-differentiated hepatocellular carcinoma (WDHCC).

[MATERIALS AND METHODS] Contrast-enhanced MRI images and clinical data of 144 pathologically confirmed HCA or WDHCC enrolled retrospectively from multiple centers between January 2015 and January 2024. Two readers reviewed images to identify imaging features and measure signal intensity on multiple phases images. The predictive model was established using binary Logistic regression, and the predictive ability was evaluated using the area under the curve (AUC), accuracy, sensitivity, and specificity by R software.

[RESULTS] Out of 144 eligible patients (35 HCAs, 109 WDHCCs), 23 in 37 indexes showed significant differences. Moreover, 10 parameters remained significant after the univariate regression analysis. To construct a highly accurate predictive model, the significant parameters were further subjected to a multivariate regression model. Six valuable factors (long axis, T1WI, T2WI/FS, capsule enhancement, septa, and cirrhosis) were selected to establish the diagnostic model. Then, a nomogram to discriminate HCA from WDHCC was built on the basis of a multivariate logistic regression model. The AUC of the MRI signal model, the clinical factors model, and the combined model in training sets and validation sets are 0.955, 0.929, 0.962, and 0.898, 0.835, 0.846, respectively. DCA and clinical impact curve was applied to assess the clinical utility of the diagnostic nomogram. Based on the DCA, the MRI signal showed superior clinical utility compared to the other models.

[CONCLUSION] MRI signal-based model provides high diagnostic performance as demonstrated in the differentiation of HCA and WDHCC, supported by a nomogram model.