Rationale and objectives: The purpose of this study was to evaluate the accuracy of contrast material-enhanced sonography in the detection of liver lesions by using an animal model.
Materials and methods: A total of 36 rabbits, 12 normal and 24 with one, two, or more VX2 tumors implanted percutaneously, were imaged on an Acuson 128XP/10 with a 7-MHz sector transducer by a sonographer blinded to the study assignments. The sonographer assigned rabbits to four groups (no, one, two, more than two tumors) based on the number of lesions detected before and then after the intravenous bolus injection of 0.5 mL of AF0150. S-VHS video segments or pre- and postcontrast images were separated, randomized, and evaluated by a blinded reader. Necropsy served as the gold standard.
Results: Classification of rabbits as normal or tumor bearing on the precontrast images produced three false-positive results and three false-negative results for the blinded sonographer and six false-positive results and two false-negative results for the blinded reader. On postcontrast images, all rabbits were correctly classified by both observers. The correlation of the classification of whether rabbits had no, one, two, or more tumors relative to the pathologic classification on precontrast images was poor to fair (K = 0.349 +/- 0.099 for the sonographer and 0.274 +/- 0.111 for the reader), whereas the postcontrast correlation was good to excellent (K = 0.924 +/- 0.099 for the sonographer and 0.809 +/- 0.076 for the reader).
Conclusion: AF0150 markedly increased the ability of the sonographer and the blinded reader to distinguish normal from tumor-bearing animals and improved the classification of rabbits with more than one liver tumor.