Objective: The purpose of our study was to test the hypothesis that, in noncirrhotic livers, large size predicts benignity of masses that homogeneously hyperenhance on arterial phase CT and then fade to isoattenuation.
Materials and methods: All multiphasic CT scans obtained at a cancer center over a 2-year period were reviewed. In consensus, three authors retrospectively identified 227 hepatic masses (> or = 5 mm) in 55 noncirrhotic patients that homogeneously hyperenhanced on arterial phase and then faded to isoattenuation: 107 masses were malignant and 120 were benign; 37 patients had benign and 18 patients had malignant masses. Two analytic approaches were pursued: per lesion and per patient. For the per-lesion analysis, the mean cross-sectional diameter of each mass was calculated and receiver operator characteristics (ROC) were assessed. For the per-patient analysis, the maximum lesion diameter was determined for each subject and logistic regression models were used to predict lesion classification (benign vs malignant) based on per-patient maximum lesion size and additional information.
Results: Masses ranged from 5 to 84.5 mm. All 29 masses > or = 22 mm were benign. Size was a statistically significant classifier of benign versus malignant lesions in the per-lesion analysis (p = 0.024, ROC area under the curve) and a significant or trend-level predictor of tumor type in the per-patient analysis (logistic regression p values of the diameter coefficients: 0.01-0.07).
Conclusion: In noncirrhotic livers, relatively large size is a significant or trend-level predictor for benign tumors. Homogeneously hyperenhancing masses > or = 22 mm that fade to isoattenuation are benign; smaller masses may be malignant.