DIY artificial pancreas system shows promise for T1D

A do-it-yourself (DIY) artificial pancreas system (APS) for Android devices seems to be a safe intervention in patients with type 1 diabetes (T1D), helping to improve glycated haemoglobin (HbA1c) levels and time in range without posing a higher risk of hypoglycaemia, according to a recent study.

“DIY systems by definition must be self-built and as for now are not regulated or approved by FDA,” the researchers said, adding that “compared to commercial hybrid closed-loop (HCL) systems, controlled clinical trials designed to demonstrate safety and effectiveness of DIY APS are missing.” The number of patients using DIY APSs are growing, underscoring the need for a better understanding of these systems.

In line with this, the researchers performed a single-centre clinical trial, enrolling 12 T1D patients (mean age 31.3±6.7 years, seven women). Participants received a DIY APS with the following elements: an insulin pump, a continuous glucose monitoring system, and an Android smartphone with the AndroidAPS application. The primary outcome was safety, defined as severe hypoglycaemia, diabetic ketoacidosis, or time spent in glycaemia <54 mg/dL.

After 12 weeks of intervention, sensor glycaemia significantly dropped from 153.3±18.0 mg/dL at baseline to 141.4±8.4 mg/dL (p<0.001). The time in range of 70–180 mg/dL also improved, jumping from 68.0±12.7 percent at baseline to 79.3±6.4 percent at the final assessment (p<0.001). [PLoS One 2021;doi:10.1371/journal.pone.0248965]

Over the same time period, the researchers documented a significant decrease in both HbA1c (6.8±0.5 percent to 6.3±0.4 percent; p<0.001) and fructosamine (351.0±31.4 to 314.6±22.3 µmol/L; p<0.001).

During the 12 weeks of intervention, participants were asked to see the research staff every 4 weeks for a control visit. These intermittent follow-ups also confirmed that there were significant improvements in average sensor glycaemia as well as in the time in the target ranges of 70–80 and 70–140 mg/dL.

Meanwhile, there was not a significant increase from baseline in terms of time in hypoglycaemia <54 mg/dL (0.25 to 0.34 percent) or <70 mg/dL (2.50 to 1.75 percent). APS likewise had no clear impact on daily insulin doses, body mass index, and body weight.

“The settings of AndroidAPS application were equal for the whole group and have not been modified through the observation,” the researchers said, however, pointing out that the settings of the insulin pump were adjusted during the control visits in accordance to patient response.

Throughout the process, they continued, the team was brave enough to uptitrate insulin doses when needed, and patients were trusting enough to not question the medical decisions too much. Eventually, the tweaks in dosing may have contributed to the good outcomes in the study.

Nevertheless, the study has crucial weaknesses, such as its observational design and small sample size. Future studies addressing these limitations, as well as employing better, more statistically powerful methodologies, would help deepen the understanding on DIY APS for T1D.

“We conclude that DIY AP systems require at the beginning a lot of commitment from patients who would like to start such treatment. However, over time, patients become accustomed to the devices, the algorithm and the effects they can achieve,” the researchers said. “It requires patience. Our results indicate that the learning process pays off with good results after some time.”