Statistical relevance of ultrasonographic criteria in the assessment of diffuse liver disease in dogs and cats

Abstract

To determine whether objectively applied ultrasonographic interpretive criteria are statistically useful in differentiating among 7 defined categories of diffuse liver disease in dogs and cats. Ultrasonographic images of 229 dogs and 104 cats. Liver parenchymal or related sonographic criteria established by the authors were retrospectively and independently applied by 3 radiologists who were not aware of patient status or patient laboratory data. Seven histologic or cytologic categories of diffuse (infiltrative but not nodular) liver diseases were jointly established by the authors and included normal liver; inflammation; round-cell neoplasia; non-round-cell infiltrative, prenodular (early) metastatic neoplasia; lipidosis; vacuolar hepatopathy; and other. Liver parenchymal sonographic criteria included parenchymal sound attenuation with increasing depth, comparative organ echogenicity (liver, spleen, and kidneys), diffuse or patchy hyperechoic or hypoechoic echotexture, uniform or coarse echotexture, portal venous clarity, and liver lobe geometry. Related extrahepatic criteria included gallbladder wall thickness, bile duct diameter, amount and character of gallbladder precipitate, nondependent shadowing in the gallbladder, hepatic vein diameter versus caudal vena cava diameter, peritoneal fluid, spleen echotexture (normal vs abnormal [characterized]), and kidney echotexture. Ultrasonographic criteria were statistically compared to the 7 categories of diffuse liver disease in search of clinically exploitable relationships. Statistical evaluation of the applied ultrasonographic criteria did not yield clinically acceptable accuracy for discrimination among the 7 categories of diffuse liver diseases (including normal liver) in either species. Criterion-based ultrasonographic appearance was insufficient to discriminate among canine and feline diffuse infiltrative liver diseases.