Oral administration of autoantigens can slow the progression of β-cell destruction in non-obese diabetic mice. We investigated whether oral administration of recombinant human insulin could protect residual β-cell function in recent-onset type 1 diabetes.

We enrolled 131 autoantibody-positive diabetic patients aged 7–40 years within 2 weeks of diagnosis (no ketoacidosis at diagnosis, weight loss <10%, polyuria for <6 weeks). They were randomly assigned 2·5 mg or 7·5 mg oral insulin daily or placebo for 1 year, in addition to subcutaneous insulin therapy. Serum C-peptide concentrations were measured in the fasting state and after stimulation, to assess β-cell function. Autoantibodies to β-cell antigens were assayed. Analyses were by intention to treat.

Baseline C-peptide and haemoglobin A_lc concentrations were similar in the three groups. During follow-up, there were no differences between the groups assigned 2·5 mg or 7·5 mg oral insulin or placebo in subcutaneous insulin requirements, haemoglobin A_lc concentrations, or measurements of fasting (mean at 12 months 0·18 [SD 0·17], 0·17 [0·17], and 0·17 [0·12] nmol/L) or stimulated C-peptide concentrations (glucagon-stimulated 0·39 [0·38], 0·37 [0·39], and 0·33 [0·24] nmol/L; meal-stimulated 0·72 [0·60], 0·49 [0·49], and 0·57 [0·51 nmol/L]. Neither age nor C-peptide concentration at entry influenced treatment effects. No differences were seen in the time-course or titres of antibodies to insulin, glutamic acid decarboxylase, or islet antigen 2.

At the doses used in this trial, oral administration of insulin initiated at clinical onset of type 1 diabetes did not prevent the deterioration of β-cell function.