Differentiating focal eosinophilic necrosis of the liver from hepatic metastases using unenhanced and portal venous phase computed tomographic imagings: Results of univariate and multivariate statistical analyses

OBJECTIVE: To evaluate the usefulness of computed tomographic (CT) predictors on portal venous phase images and a new CT predictor on unenhanced images in differentiating focal eosinophilic necrosis (FEN) from hepatic metastases. MATERIALS AND METHODS: Computed tomographic findings were analyzed in 20 patients with FEN (n = 84) and 23 patients with hepatic metastases (n = 81). Computed tomographic features were compared using univariate and multivariate analyses to determine significant findings. RESULTS: An irregular shape, a subtle low attenuation, a fuzzy margin, and absences of a contour bulging and a rimlike enhancement on the portal venous phase images and absence of a discernable low attenuation on the unenhanced images were significant variables favoring FEN in univariate analysis. In multivariate analysis, no discernable low attenuation on the unenhanced images and an irregular shape and a subtle low attenuation on the portal venous phase images were the variables independently favoring FEN. CONCLUSIONS: Not only portal phase images but also unenhanced images can be helpful to differentiate FEN from hepatic metastases.