RESUMO
AIMS: Anti-insulin antibodies in insulin-treated diabetes can derange glycaemia, but are under-recognised. Detection of significant antibodies is complicated by antigenically distinct insulin analogues. We evaluated a pragmatic biochemical approach to identifying actionable antibodies, and assessed its utility in therapeutic decision making. METHODS: Forty people with insulin-treated diabetes and combinations of insulin resistance, nocturnal/matutinal hypoglycaemia, and unexplained ketoacidosis were studied using broad-specificity insulin immunoassays, polyethylene glycol (PEG) precipitation and gel filtration chromatography (GFC) with or without ex vivo insulin preincubation. RESULTS: Twenty-seven people had insulin immunoreactivity (IIR) below 3000 pmol/L that fell less than 50% after PEG precipitation. Insulin binding by antibodies in this group was low and judged insignificant. In 8 people IIR was above 3000 pmol/L and fell by more than 50% after PEG precipitation. GFC demonstrated substantial high molecular weight (HMW) IIR in 7 of these 8. In this group antibodies were judged likely significant. In 2 people immunosuppression was introduced, with a good clinical result in one but only a biochemical response in another. In 6 people adjustment of insulin delivery was subsequently informed by knowledge of underlying antibody. In a final group of 5 participants IIR was below 3000 pmol/L but fell by more than 50% after PEG precipitation. In 4 of these GFC demonstrated low levels of HMW IIR and antibody significance was judged indeterminate. CONCLUSIONS: Anti-insulin antibodies should be considered in insulin-treated diabetes with unexplained glycaemic lability. Combining immunoassays with PEG precipitation can stratify their significance. Antibody depletion may be beneficial, but conservative measures often suffice.