Can one mix two non-specific antisera and obtain a specific radioimmunoassay?

Sekihara et al. have proposed that it is possible to combine two antisera, each of which is unsatisfactory for a clinical radioimmunoassay due to cross-reactivity problems, and obtain an assay which is of sufficiently good specificity for practical application. The present report provides a theoretical analysis of this problem in a "reduced" case of minimal complexity. We assume infinitesimal concentration of tracer, equilibrium of reactants, perfect separation of bound and free, and that each of the two antisera contain only a single class of antibody sites which can bind to the desired ligand or to a cross-reacting species. Numerical methods are used to generate "ideal" dose response curves. The specificity is evaluated by three criteria: 1) as the ratio of ligand concentrations resulting in 10 or 50% reduction of binding of labeled ligand to antibody, i.e. %B/BO = 90 or 50%; 2) the %B/BO or %B/T at an arbitrary dose level; or 3) the apparent amount of ligand present, for an arbitrary dose of crossreacting ligand. Results indicate that mixing of two nonspecific antisera (each cross-reacting with a different ligand) results in a radioimmunoassay system with a specificity intermediate between that obtained with either of the antisera used alone. Whether this will provide a "satisfactory" assay depends on the purposes for which it is intended, the expected concentrations of cross-reacting ligands, etc. Computer simulation studies may be utilized to select the optimal ratio of the two antisera being used for the assay.