The insulin ball

In August, 2005, a 62-year-old Japanese man, with a 20-year history of type 1 diabetes, was admitted to our hospital, for investigation of apparent insulin resistance. Since January, 2004, he had been taking insulin in a longacting (glargine) and a shortacting (lispro) form, at a total of 66 units a day. Despite the dose being gradually increased to 94 units a day, his blood glucose concen trations had subsequently been high, and poorly con trolled. Examination showed nothing of note. Blood tests gave a value for glycosylated haemoglobin (HbA1c) of 8∙6% (target value ≤6·5%). The patient’s body-mass index was 21∙1 kg/m2 (weight 57∙5 kg). Although we kept the patient on a strict diet, and increased the total daily dose of insulin to 116 units (lispro 94; glargine 22), his blood glucose concentration, before meals, remained 7∙8–21∙1 mmol/L. Blood tests showed normal concentrations of corticotropin, cortisol, thyroid hormones, glucagon, growth hormone, and carcino-embryonic antigen, and undetectable quantities of antibodies to insulin; urinary concen trations of meta nephrines were normal. Ultrasonography of the abdomen showed nothing of note. However, re-examination of the lower abdomen revealed two lumps, one on each side, under the skin; each contained a hard, irregular nodule, 3–4 cm in diameter. The patient had noticed the lumps several months previously; he routinely injected insulin into the nodules, since they were easily grasped, and injection was less painful than elsewhere. We admin istered insulin via a diff erent part of the abdomen; hypoglycaemia ensued. We rapidly decreased insulin doses. Good blood glucose concentrations were obtained with a total daily dose of 24 units (lispro 18; glargine 6). Abdominal MRI showed the nodules to lie in the fat under the skin; unlike fat, they had low signal intensity on T1-weighted (fi gure) and T2-weighted images. We took a biopsy sample of a nodule; histopathological analysis showed it to consist largely of amorphous, acellular, eosinophilic material. When the material was stained with Congo red, and seen with polarised light, we saw green birefringence, diagnostic of amyloid. On immunohistochemical testing, the amyloid deposit was positively stained by monoclonal antibodies to human insulin. When last seen, in November, 2008, the patient needed 66 units a day of insulin—injected outside the lumps, which were smaller than before. Common causes of insulin resistance include obesity; sever physical stress; prescribed drugs, such as glucocorticoids; and endocrine disorders, such as Cushing’s syndrome, phaeochromocytoma, or acromegaly. Rarer causes include mutations of the insulin receptor gene; antibodies to insulin, or to the insulin receptor; and lipodystrophy syndromes. Doctors should always consider if, and how, the patient is injecting insulin. Amyloid deposits consist of proteins arranged in a cross-β conformation: pathological amyloid deposits can be formed of various proteins, including α-synuclein in Parkinson’s disease, and prions in Creutzfeldt-Jakob disease. For amyloid to consist of insulin is rare. In all reported cases of insulin-induced amyloidosis, of which we are aware, the patient took non-human insulins: we speculate that aminoacid diff erences between endogenous and injected insulin may have contributed to the development of amyloidosis. Partly because use of nonhuman insulin is increasing, and partly because we have seen four cases recently, we speculate that insulininduced amyloidosis may be under diagnosed: for instance, by being mistaken for lipo hypertrophy. Injection into fatty lumps can reduce the eff ectiveness of insulin, though less markedly than does injection into amyloid. Amyloid lumps are typically harder, and more discrete, than fatty lumps of lipo hyper trophy; MRI can assist with diagnosis. We refer to amyloid lumps of insulin as “insulin balls”.