Predictive factors for prolonged remission after autologous hematopoietic stem cell transplantation in young patients with type 1 diabetes mellitus

Background aims. Autologous hematopoietic stem cell transplantation (auto-HSCT) followed by immunoablation is a promising therapy for type 1 diabetes mellitus (T1DM) treatment due to the immunosuppression and immunomodulation mechanisms. Indeed, a considerable number of patients have been able to discontinue insulin use with this treatment. However, nonresponse and relapse occur after auto-HSCT. It is important to select the patients who can potentially benefit from this treatment, but the factors that might influence the therapeutic outcome are unclear. The objective of this study was to explore the predictors for prolonged remission after auto-HSCT therapy. Methods. The data for this study were extracted from an open-label prospective study, which was performed to treat new-onset T1DM patients with auto-HSCT. The 128 patients were categorized into insulin-free (IF) or insulin-dependent (ID) groups according to their response to treatment during the follow-up. We compared the baseline data of the two groups and explored possible prognostic factors and their odd ratios (ORs) with univariate analysis and multivariate logistic regression. Receiver operating characteristic curves (ROC) were performed to test the model discrimination function. Results. During a follow-up of 28.5 +/- 8.3 months, 71 of 128 patients in the IF group discontinued insulin use, whereas 57 of 128 patients in the ID group did not decrease their insulin dose or resumed insulin treatment after a transient remission. Multivariate logistic regression analysis demonstrated that prolonged remission was positively correlated with fasting C-peptide level (OR = 2.60, 95% confidence interval [CI]: 1.16-5.85) but negatively correlated with onset age (OR = 0.36, 95% CI: 0.14-0.88) and tumor necrosis factor-alpha levels (OR = 0.32, 95% CI: 0.14-0.73). ROC analysis confirmed the combined predictive function of these three variables (AUC = 0.739, 95% CI: 0.655-0.824). Conclusions. Age and fasting C-peptide and tumor necrosis factor-alpha levels were identified as possible predictors for prolonged remission following auto-HSCT therapy.