A simple and precise equilibrium binding assay has been developed for the quantitation of circulating insulin antibodies. This procedure differs from previously reported methods in that (1) serum samples were deinsulinized in order to avoid endogenous insulin's interfering with the binding assay, (2) l25I-monoiodoinsulin was used to prevent artifacts resulting from variability in ligand binding due to excessive iodination, and (3) separation of free and bound insulin was accomplished by rapid precipitation of the hormone-antibody complexes without disturbing the binding equilibrium. This approach prevented dissociation and underestimation of the bound species.
In agreement with previous reports, two classes of antibody sites (high-affinity-low-capacity and low-affinity-high-capacity) were demonstrated in the sera of insulin-treated diabetic patients. No consistent differences existed in the association constants and/or binding capacities for insulins of bovine, porcine, and human origin, even though the immunogenic stimulus for antibody formation in these patients was treatment with commercial preparations of bovine and porcine insulins. Thus, higher affinities and/or binding capacities for human versus bovine and porcine insulins does not provide evidence for an autoimmune origin of insulin antibodies, as has been previously proposed. Significant differences in the antibody-binding capacities for all three species of insulin were found in individual serum samples, and underestimation of antibody titers may occur if only one species of the hormone is used.
Kinetic studies on the dissociation of the insulin-antibody complexes revealed two classes of dissociating sites, in agreement with the equilibrium binding data. The corresponding dissociation rate constants were large enough to implicate the release of insulin from the pool of antibody-bound hormone as a possible hypoglycemic mechanism in some diabetic patients.