Insulin antibodies and hypoglycemia in diabetic patients. Can a quantitative analysis of antibody binding predict the risk of hypoglycemia?

We report a noninsulin-dependent diabetes mellitus (NIDDM) patient with spontaneous, severe hypoglycemic reactions and the presence of insulin antibodies. He had a remote antecedent history of beef-pork insulin therapy as well as exposure to hydralazine. Detailed insulin binding kinetic studies were performed in this patient as well as in six other insulin-treated diabetic patients with anti-insulin antibodies (three with and three without an obvious cause of hypoglycemia). Sera from the current patient and five of the six other diabetic patients (one NIDDM, four IDDM) revealed two types of binding sites: high-affinity with low capacity (Kd, 0.4-12.4 x 10(-9) mol/L; binding capacity, 0.6-659 mU/L) and low-affinity with high capacity (Kd, 0.3 to 35.7 x 10(-7) mol/L; binding capacity; 202-113,680 mU/L). One NIDDM patient had only high-affinity antibodies (Kd, 22.9 x 10(-9) mol/L; binding capacity of 78 mU/L). Type of diabetes mellitus, insulin antibody titers or their binding capacities, insulin levels (total, bound, or free), and bioavailable insulin were not related to hypoglycemic reactions. Two calculated values by the method described tended to discriminate patients with and without hypoglycemia. The calculated amount of low-affinity antibody bound insulin ranged from 69.4-2090 mU/L vs < 4-70.6 mU/L in patients with and without hypoglycemia, respectively. The best discrimination was afford by the percent saturation of low-affinity binding sites; values were clearly higher in the patients with hypoglycemia (2.5-34.4%) than in those without hypoglycemia (not detectable, 0.06, 0.15%). Consideration of the possible drug-associated insulin antibody formation in insulin-treated diabetics and the novel quantitative analysis of the antibody binding kinetics should prove helpful in evaluating patients with high insulin antibody titers and assessing the risk of hypoglycemia.