Reproducibility of Liver Biopsy Diagnosis in Relation to the Size of the Specimen

Abstract

For the purpose of evaluating the reproducibility of the chief diagnosis and several individual lesions in liver needle biopsies in relation to the size of the biopsy, we investigated without knowledge of clinical data 100 liver specimens in five sessions with an increasing amount of tissue made visible. The material consisted of routine liver biopsy specimens that were 25 mm or longer, comprising 50 with acute viral hepatitis (AVH), 10 with chronic aggressive hepatitis (CAH), 10 with micronodular cirrhosis, 10 with macronodular cirrhosis, 10 with steatosis, and 10 with other diagnoses. For the statistical analysis of the chief diagnoses, Friedman's two-way analysis of variance and Cochran's Q-test have been used. The predictive value of a lesion present and a lesion absent was used in evaluating the efficiency of the diagnoses of the individual lesions. The microscopic diagnosis of AVH was found to be safe even on biopsies only 5 mm long. Furthermore, the predictive value of both presence and absence of various individual lesions (piecemeal necrosis, confluent necrosis, etc.) in AVH was higher in long biopsies than in short. The evaluation of such individual lesions was shown to be safer when the biopsy was 15 mm or longer. Thus a clinically acceptable accuracy in the histological diagnosis requires 15 mm. The diagnostic reproducibility of the chief diagnosis of CAH and cirrhosis increases with increasing amount of available tissue. A considerable number of biopsies less than 25 mm long are not diagnostic.

Key-words:

• Liver needle biopsies
• predictive values
• reproducibility