Comparison of pathologic and normal sera by immune complex determination: five disease groups within 190 samples are discriminated by computer-selected combinations of 13 methods. Report of the Italian committee for the study of immune complexes (WIC).

Pathological (190) and normal (33) sera were tested for their content of circulating immune complexes (CIC) by a battery of 13 assays performed in 11 laboratories. Statistical processing was done both by pooling all pathological samples and by extracting those falling into well-defined disease groups, i.e., rheumatoid arthritis, diabetes, lupus, melanoma, and glomerulonephritis. Highly significant correlations between methods--taken two at a time--for each disease differed in proportion (ranging from 6 to 30%) and in the pattern displayed on a checkerboard. Disease-linked patterns were also found when a function maximizing discrimination between pathological and normal samples was derived by combining the information from all methods. Here the order and the weight attributed by the computer to the methods differed for each of the disease groups. Taken together these results are interpreted as an indication that all assays may not determine the same classes of CIC, and thus vary in sensitivity depending on the prevailing properties of the complexes present in the serum, which in turn may depend on the etiology, pathogenesis, and stage of the disease.