Introduction/Purpose: With the increased use of abdominal CT and ultrasonography, hepatic masses of uncertain clinical relevance are often identified. Determining the benign or malignant nature of these masses can drastically alter patient management but remains a diagnostic challenge without invasive intervention. The objectives of this study were to determine the statistical relevance of quantitative and qualitative evaluation criteria of hepatic masses using triphasic CT, B-mode, color flow and power Doppler ultrasonography, and to compare the diagnostic accuracy of these modalities for predicting mass type classified by histopathology obtained from ultrasound guided tru- cut biopsies, surgical resection, or necropsy.
Methods: Forty-four dogs with a hepatic mass identified on triphasic CT were enrolled prospectively and underwent hepatic sonography followed by histologic evaluation of the mass (obtained by tru-cut biopsy, surgical mass excision or necropsy). For each mass, 5 radiologists, blinded to the histopathology, assessed 57 tomographic and 13 sonographic qualitative and quantitative variables independently. Discriminant analysis of the variables was performed using the Wilks’ Lambda method, and variables with an F value < 0.05 were included for a stepwise analysis. At any step those variables with an F value > 0.1 were removed. Fisher’s linear discriminant equations were used to produce an equation-based prediction of evaluation criteria and the relationship to histopathologic diagnosis for those variables with a statistical significance.
Results: An equation including the lowest delayed-phase absolute enhancement of the mass (post-contrast attenuation – pre-contrast attenuation) and the highest venous- phase mass conspicuity (mass attenuation – liver attenuation) correctly classified 93.5% of the liver masses as benign or malignant. Of these two CT variables, only the lowest delayed-phase absolute enhancement of the mass retained significance independently and correctly classified 91.3% of the masses, using a cut-off of 37 Hounsfield Units as determined by Fisher’s linear discriminant analysis. Masses with a low delayed-phase absolute enhancement greater than 37 HU were classified as benign and masses with a low delayed-phase absolute enhancement less than 37 HU were classified as malignant, with a sensitivity of 93.3% and specificity of 87.5% for malignancy. The only ultrasound variable retaining significance in the stepwise analysis was the subjective presence or absence of cavitations within the mass. Categorizing masses with cavitations as malignant and those without cavitations as benign resulted in a diagnostic accuracy of 80.4%.
Discussion/Conclusion: Triphasic CT had a higher accuracy in predicting lesion classification than B-mode, color or power Doppler ultrasonography. The lowest delayed- phase absolute enhancement of the mass is a simple calculation requiring 2 measurements and aids in the classification of benign and malignant liver masses.