ABSTRACT

We aimed to establish a liver function evaluation model by combining multiparametric magnetic resonance imaging (MRI) with liver volume (LV) and further verify the effectiveness of the model to evaluate liver function.

This retrospective study included 101 consecutive cirrhosis patients (69 cases for modeling group and 32 cases for validation group) who underwent gadoxetic acid-enhanced MRI. Five signal intensity parameters were obtained by measuring the signal intensities of the liver, spleen, and erector spinae before and 20 minutes after gadoxetic acid disodium enhancement. The diffusion coefficient (D), pseudo-diffusion coefficient (D*), and perfusion fraction (f) were obtained from intravoxel incoherent motion diffusion-weighted imaging. The LV parameters (Vliver, Vspleen, and Vliver/Vspleen) were obtained using 3-dimensional image generation software. The most effective parameter was selected from each of the 3 methods, and a multivariate regression model for liver function evaluation was established and validated.

In the modeling group, relative enhancement (RE), D*, and Vliver/Vspleen showed significant differences among the different liver function groups (P < .001). Receiver operating characteristic analysis showed that these parameters had the highest area under the curve (AUC) values for distinguishing Child-Pugh A from Child-Pugh B and C groups (0.917, 0.929, and 0.885, respectively). The following liver function model was obtained by multivariate regression analysis: F(x)=3.96 − 1.243 (RE) − 0.034 (D*) − 0.080 (Vliver/Vspleen) (R2=0.811, P < .001). In the patients with cirrhosis, the F(x) of Child-Pugh A, B, and C were 1.16 ± 0.44, 1.95 ± 0.29, and 2.79 ± 0.38, respectively. In the validation group, the AUC for F(x) to distinguish Child-Pugh A from Child-Pugh B and C was 0.973.

Combining multiparametric MRI with LV effectively distinguished patients with different ChildPugh grades. This model could hence be useful as a novel radiological marker to estimate the liver function.