Diagnosis of liver metastases: Can diffusion-weighted imaging (DWI) be used as a stand alone sequence?

Abstract

To evaluate if diffusion-weighted MRI (DWI) can replace gadolinium-enhanced MRI (Gd-MRI) for diagnosing liver metastases. The diagnostic accuracy of both techniques alone and in combination are compared. Sixty-eight patients with histologically proven primary extrahepatic tumors were included in this retrospective study. Lesions included 62 metastases and 130 benign lesions. Three image sets (unenhanced T1 and T2/gadolinium enhanced T1 (Gd-MRI), DWI and combination of both) were reviewed independently by 3 observers. The areas under the receiver operating characteristic curves (A(z)), sensitivity and specificity for the 3 image sets were compared. The standard of reference was either histopathology or multi-modality and clinical follow-up. Pooled data showed higher diagnostic accuracy for the combined set (A(z)=0.93) compared to Gd-MRI (p=0.001) and DWI (p<0.0001). No difference was found between the performance of Gd-MRI and DWI (p=0.09). Sensitivity for the combined set was higher than Gd-MRI (p=0.0003) and DWI (p=0.0034). Specificity for DWI was lower than Gd-MRI (p<0.0001) and the combined set (p<0.0001). The diagnostic performance of DWI is equal to that of Gd-MRI. DWI alone can be used in patients where gadolinium contrast administration is not allowed. Combination of Gd-MRI and DWI significantly increases diagnostic accuracy.