The authors Castaldo et al. (Clin. Chem., 1994, 30: 478-83) state, that the ascitic lactate dehydrogenase and ascitic cholesterol association correctly identified 100% of malignant ascites from ascites associated with cirrhosis and/or hepatocellular carcinoma, with help of stepwise multiple linear discriminant analysis. The free software Capsules--Ascites is via internet available (http:¿www.leeds.ac.uk/acb), which use the mathematical formula from this article. As we argue, Castaldo's state is not correct. Three independent multidimensional statistical methods--bivariate reference regions (program EVAL-KIT), cluster analysis (program BioAnalyst), geometrical distance classification (program GEODICLA) applicated on Castaldo's original data showed that lactate dehydrogenase and cholesterol have not satisfactory absolute discriminative power between malignant from nonmalignant ascites in general, but the probability to determinate the right diagnosis is about 91-93%. Research conducted in correctly selected probands should provide information which is valid not only for the selected sample but for the entire population, to achieve more generally valid conclusions, useful for practical decisions. In addition, in the compiled table show we sensitivity and specificity of different laboratory parameters publicated in 11 original articles in the nineties, all with diagnostic efficiency less 100%.

Keywords: Ascites /etiology/; Ascitic Fluid /chemistry/; Carcinoma, Hepatocellular /complications/; Cholesterol /analysis/; Diagnosis, Differential; Humans; L-Lactate Dehydrogenase /analysis/; Liver Cirrhosis /complications/; Liver Neoplasms /complications/; Sensitivity and Specificity