We evaluated the production of antibodies against insulin in a genetically well-defined population. In the first study, 124 young patients with type I diabetes for longer than 6 mo were included. Anti-insulin antibodies were detected by polyethyleneglycol (PEG) precipitation after incubation of acidified, charcoal-stripped sera with 125I-labeled pork insulin andwere expressed as microunits insulin bound per milliliter whole serum. For comparison, the patients were divided into six groups based on HLA DR antigens: 3/3, 3 /−, 4/4, 4 /− , 3/4, and − / − (– is non-DR3 or -DR4). The mean age of the patients was 14.7 ± 0.5 yr; the duration of diabetes was 5.8 ± 0.4 yr; and the glucose control, as measured byhemoglobin A1c was average (7.6 ± 0.2%). There were no significant differences in any of these parameters among the patients in any of the HLA DR groups. Patients expressing DR3/3 had significantly lower insulin binding than the rest of the groups (2.5 ± 0.4 vs. 13.6 ± 1.4 μU/ml, P < 0.0001). Patients with DR3/− did not differ in insulin-binding capacity from the other groups. The type of insulin used for replacement was not correlated with the serum insulin-binding capacity. In a second study, sera from 48 children, newlydiagnosed with type I diabetes, were examined for the presence of insulin binding before treatment with exogenous insulin and compared with sera from 80 children without diabetes or a family history of diabetes and from 103 unaffected HLA-identical or haploidentical siblings of a child withtype I diabetes. The sera from the newly diagnosed diabetics bound significantly more insulin in the assay (1.02 ± 0.06 μUsol;ml than the sera from either of the two control groups (0.52 ± 0.01 and 0.51 ± 0.02%, respectively, P = 0.0012). Sera from the newly diagnosed girls bound significantly more insulin than did those from the boys (1.13 ± 0.09 vs. 0.89 ± 0.07%, P < 0.02), and there was a highly significant (P = 0.0017) negative correlation between the age at onset of diabetes and the insulin bound by serum before insulin treatment.
Finally, sera from six siblings of type I diabetics who became diabetic themselves were obtained when these children were metabolically normal and again just before insulin treatment was begun. Four of these children had a significant increase in their serum insulin binding from normal values to clearly elevated levels over intervals before metabolic decompensation.
These data support the following conclusions. The reduced insulin binding previously observed in type I diabetics expressing HLA DR3 can be entirely attributed to the very low binding in individuals homozygous for this antigen. Thus, human antibody response to exogenous insulin is probably regulated by a dominant gene in negative linkage disequilibrium with HLA DR3. Most younger patients have detectable insulin binding at diagnosis of type I diabetes. The level of binding is reduced in older patients, suggesting either that younger children have a better response to insulin released from injured β-cells in a form that is antigenic or that younger children have animmaturity in the clearance of antigen-antibody complexes. Our preliminary data suggest that the elevation in insulin binding is a late event in the pathogenesis of type I diabetes, occurring soon before the symptoms are present.