Probability of remission in individual in early adult insulin dependent diabetic patients. Results from the Cyclosporine Diabetes French Study Group.

The aim of this study was to determine which candidates were suitable for immunotherapy among adult insulin dependent diabetic patients of recent onset. A statistical analysis was performed using the results of a multicentre randomized trial of cyclosporine versus placebo after nine months of treatment. When the baseline characteristics of the patients in remission were compared with those not in remission, there was no difference observed either in initial residual beta-cell function (glucagon stimulated C-peptide level), or in immunological markers (T4 and T8 lymphocytes counts, Interleukin 2). The parameters showing the most difference were, in addition to treatment group, the duration of diabetes symptoms and body mass index at inclusion, and the HLA-DR phenotype. This was confirmed using a logistic regression analysis, in which these variables were found to be significantly related to remission. The probability of remission in each individual patient was then calculated using these variables in the mathematical function provided by the logistic model. Ninety eight out of 110 patients were correctly classified using this method. In addition, it must be noted that only subjects adequately treated by cyclosporine were still in complete remission after a one year follow-up. Conversely, it appeared that immunosuppression in subjects having a predicted probability of remission lower than 0.35 using the mathematical function, and being non-DR3, non-DR4 has to be avoided. These results will be useful in optimizing the recruitment of patients in on-going or future trials of immunotherapy in early diabetes.