Type I diabetes mellitus is an autoimmune disease in which destruction of pancreatic islet cells leads to loss of insulin secretion. The process is thought to be mediated by T cells, including CD4⁺ "helper" cells [1]. We describe a patient whose illness may shed light on whether human immunodeficiency virus (HIV) infection, which destroys CD4⁺ cells, modulates the clinical course of diabetes mellitus [2, 3].

A 46-year-old man was diagnosed with diabetes mellitus in July 1990, when he was found to have hyperglycemia and ketoacidosis. Insulin therapy (approximately 0.8 U/kg per day) stabilized his blood glucose at mildly elevated levels. In July 1991, he was diagnosed with the acquired immunodeficiency syndrome (AIDS) in association with toxoplasmosis of the central nervous system and was treated with pyrimethamine-sulfadiazine. He also received zidovudine and dexamethasone from July to September. In October, after 15 months of insulin treatment and 3 months after discontinuing dexamethasone therapy, he noted decreasing insulin requirements and soon discontinued all insulin use; he had a normal glycosylated hemoglobin level of 5.8%. His CD4 cell count at that time was very low (29 cells/mm³). His weight had been stable (79.5 kg; height, 183 cm), and thyroid function tests and serum cortisol levels were normal. Islet cell antibodies, tested for the first time, were positive. Within a few months, glucose values increased, requiring resumption of insulin therapy, although at lower doses (<0.2 U/kg per day). A random C-peptide level at that time was 2.8 µg/L (normal, 0.3 to 3.7 µg/L).

Hyperglycemia with ketoacidosis has been reported in patients with AIDS because of pentamidine-induced pancreatitis [4]. A previous report [2] of the concurrent development of diabetes and AIDS did not mention pentamidine use. Our patient became diabetic before developing AIDS, had islet cell antibodies (consistent with autoimmune diabetes), and never received pentamidine. Some patients with diabetes mellitus have temporary clinical remissions (the "honeymoon" period), but this usually occurs much sooner after the diagnosis [5]. We hypothesize that the patient developed HIV-related immune suppression at a time when pancreatic ß-cell destruction was profound but not complete and that the loss of effective CD4⁺ cell functioning temporarily arrested the autoimmune insulitis, leading to a clinical remission of his diabetes.