The artificial pancreas in children: First tests in Italy

Background. The artificial pancreas (AP) system based on the Modular Model Predictive Control algorithm (MMPC) running on the wearable platform Diabetes Assistant (DiAs) has been used in adult outpatients with type 1 diabetes (T1D), reducing the time spent in hypoglycemia and increasing the time spent in the near-normal range of 70-180 mg/dl compared to Sensor Augmented Pump therapy (SAP). Objective. In the PEDiatric ARtificial PANcreas (PedArPan) project we tested a children-specific version of the MMPC on DiAs in 5-9 yearold children during a summer camp. This is the first outpatient AP trial in a population of this age. Study design and methods. Thirty 5-9 year-old children with T1D and their parents were recruited from five Italian pediatric centers and completed an open label, randomized, crossover trial in a summer camp (supervised outpatient setting). Glucose control in three days of AP use was compared with three days of parent-managed SAP. Results. Overnight (00:00-07:30). The AP significantly reduced time-in-hypo compared to SAP, 0% [0.00-2.22] vs. 2.19% [0.00- 12.30], p = 0.002, with no significant difference in time-in-target (70-180 mg/dl), 55.97% (22.53) vs. 59.69% (21.24), p = 0.430, or time-in-tight-target (80-140 mg/dl), 31.27% (20.23) vs. 33.02% (19.80), p = 0.694. Overall (00:00-24:00). There was a threefold reduction of time-in-hypo, 2.02% [1.17-4.54] vs. 6.67% [2.26-11.54], p < 0.001, with AP at the expense of some deterioration of time-in-target, 56.79% (13.47) vs. 63.09% (10.98), p = 0.022. Conclusions. This study showed the feasibility and safety of a wearable MMPC in young children. Next steps will include overnight testing at home and re-tuning the algorithm to boost efficacy during daytime.