The Clinical Significance of Classifications of Cirrhosis. A Comparison between Conventional Criteria and Numerical Taxonomy

Abstract

Classification of patients with cirrhosis serves the clinical purpose of selecting patients for adequate therapy. A material of 441 patients from a randomized clinical trial of prednisone treatment were classified according to different criteria, and the effect of this therapy was assessed by comparing the survival curves of controls and treated patients in the groups. Six classifications based on single criteria were studied, viz. females vs. males, alcoholics vs. non-alcoholics, compensated vs. decompensated cases, active vs. inactive cases, manifest vs. latent cases, and non-primary biliary vs. primary biliary cirrhosis. The latter group was too small for further analysis. Furthermore, the material was classified by numerical taxonomy, based on 56 clinical symptoms and signs and histological and serological changes. The taxonomical classification showed interdependence with all the clinical classifications, except that of primary biliary cirrhosis. Three clinical classifications, viz. those based on sex, alcoholism, and compensation, and one taxonomic class showed a significant therapeutic relevance. Combination of these criteria revealed that female patients with compensated non-alcoholic cirrhosis, previously demonstrated to benefit from prednisone treatment, showed a markedly greater prednisone effect if they belonged to the taxonomic group than if they did not. According to clinical, histologic, and laboratory data, the latter group was characterized as less severe cases than the former one. Whereas this discrepancy between clinical severity and therapeutic effect cannot be fully explained, the data do suggest that clinical criteria may not be ideal for selection of patients with cirrhosis for steroid therapy, and that numerical taxonomy may provide clues for more clinically significant classifications.