A 21-year-old white woman (BMI 21.2 kg/m2) was admitted for management of uncontrolled diabetes with cutaneous allergies to insulin. Past medical history was marked by several allergies, including coconuts and penicillin with laryngeal edema.
Type 1 diabetes was diagnosed 4 years previously and treated by three daily injections of semisynthetic human insulin (48 units/day). Four months later, the patient developed a local allergic reaction, a nettle rash without systemic manifestation that involved all injection sites. This reaction began <5 min after each injection despite H1 antihistamine treatment and subsided after 3–4 h, suggesting a type 1 IgE-mediated hypersensitivity reaction. She had a significant eosinophilia at 800.106/l (normal 0–500). Prick skin testing was negative but intradermal tests with animal or human insulin, NPH or regular, and protamine were both positive without dilution. Unfortunately, rapid-acting analogs of insulin were not tested, but the patient developed a cutaneous reaction after a premeal injection of lispro (Eli Lilly), which is a recombinant analog of insulin that may be less antigenic because it does not aggregate to form polymers (1,2). Poor compliance resulted in intermittent insulin administration and poor metabolic control. HbA1c was 13.5% (normal <6%) at entry.
Gradual desensitization with low doses of insulin was not appropriate because of the subject’s strict insulin requirements. Based on the few literature reports available (3–5), we initiated a treatment with continuous subcutaneous insulin using lispro insulin at a basal rate of 1.6 units/h. We chose to use an external insulin pump infusion as a low-dose provider for both desensitization and treatment of diabetes. Boluses were replaced with temporarily increased basal rates (2 units/h) over 3 h starting 1 h before meals, which were based on low- glycemic index foods (6), to avoid potential allergy reactivation by the necessarily large premeal doses of insulin. The Quickset infusion set (MiniMed) was used because there was no need for additional adhesive. The usual antihistamine oral treatment (cetirizine) was maintained.
Since the beginning of constant lispro infusion, we have not observed any local reaction at the insertion site of the catheter or elsewhere. The patient’s glycemic profile improved significantly. HbA1c, which was initially at 13.5%, was reduced to 8.2% after 3 months and remained between 7.5 and 8% during follow-up. The average capillary blood glucose values over the last month were 5.66 ± 1.65 mmol/l premeal and 8.25 ± 1.10 mmol/l 2-h postmeal. She reported less than two minor hypoglycemic episodes every week, and two severe episodes occurred because of physical activity. A hyperglycemic episode without ketosis that followed transient corticosteroid therapy to treat an allergic reaction to a wasp sting was successfully treated with temporarily increased basal rates (3.5 units/h) of insulin.
Although our patient developed an allergy to the insulin molecule itself, she was successfully treated using continuous subcutaneous infusion of lispro insulin with only an external insulin pump. One year later, although intradermal tests remained positive, particularly with rapid-acting insulin analogs, we could stop antihistamine treatment and introduce premeal boluses (<8 units) without reactivating cutaneous allergies.
