Management of Pregnancy in a Patient with Type 1 Diabetes and Hypoglycemia Unawareness by DBL-hu Closed-Loop Insulin Delivery System

Data regarding the performance of available hybrid closed-loop systems during pregnancy in type 1 diabetes (T1D) are limited. We report data of patient here, with highly unstable T1D with two severe unaware hypoglycemias per year with a Gold score at 4. Islet transplantation had not been retained. The patient has been educated to the carbohydrate counting, insulin/carb ratios, hypoglycemia recognition, and she was seen every month during pregnancy to verify that all was corrected. Analysis of time in target range, time in hyperglycemia, and time in hypoglycemia is reported 5 weeks before pregnancy, from early pregnancy to 12 weeks of gestation (WG), from 13 to 24 WG, from 25 to 32 WG, from 33 to 38 WG, and during breastfeeding.

The closed-loop Diabeloop for highly unstable diabetes (DBL-hu) system allows several customization settings, including the total daily insulin dose (TDD). We observed that higher insulin needs were paralleled with the requirement to increase the tuned TDD value, from an initial setting of 50 IU to a topping value of 73 IU. The insulin requirements were reduced by almost twofold at delivery and during breastfeeding at 37 IU/day, then steadily increased to 44 IU/day when breastfeeding was stopped.

As shown in Figure 1, the percentage of time in target range, starting at 25.96% in preconception, improved to 28.84%, 49.02%, 57.24%, and 80.33% for the four next periods, respectively. The glucose management index also improved reaching 5.54% between 33 and 38 WG. The percentage of time in hypoglycemia (<63 mg/dL) was kept <0.28% during pregnancy, and <0.14% during all periods except 13–24 WG. Coefficient of variation improved from 26.48% ± 4.03% to 19.71% ± 3.08% in late pregnancy.

OPEN IN VIEWER

FIG. 1.

Evolution of metabolic parameters/metrics during pregnancy.

Both daytime and nighttime periods in target range were steadily improved, topping both >75% at 13 WG till delivery at 82.10% ± 9.92% for the daytime and 77.14% ± 14.39% for the nighttime. Time in hypoglycemia (<54 mg/dL) were 0.05% (±0.07) in early pregnancy WG 1–12, 0.11% (±0.17) to 0.02% (±0.05) for WG 13–24 to WG 25–32, then canceled during WG 33–38 at 0.0% while achieving highest time in target range; breastfeeding period was 0.01% (±0.03), identical to postbreastfeeding period but higher than prepregnancy period (0.0%). Regarding the fetal outcome, a baby with a birth weight of 4070 g was born at 38 ± 1 WG. Despite metabolic optimization, macrosomia may be related to the delay in metabolic management in early pregnancy.

We have reported that management of highly unstable T1D in patients experiencing severe hypoglycemic crisis was feasible with a dedicated closed-loop insulin delivery system. ¹ DBL-hu is a hybrid CL insulin delivery combining a Dexcom G6 CGM device, a Kaleido insulin pump and the investigational DBL-hu software into a dedicated controller handset. In brief, the algorithm is a machine learning system, allowing customization through 45 settings to respond to the diversity of existing T1D pathophysiological specificities. ¹

We report that pregnancy, considered as a situation with further glycemic instability, ² can be successfully managed with the same closed-loop treatment. The use of this device was accompanied by a clear optimization of metabolic indicators especially in the third trimester. We confirm that total insulin requirement increased during pregnancy. ³ In a report comparing day-and-night HCL to sensor-augmented pump therapy among 16 T1D pregnant women, Stewart et al. observed comparable percentage of time in range, mean glucose values, and proportions of time spent >140 mg/dL with both treatments. ⁴ We emphasize the fact that we used a customizable system allowing to fine-tune the reactivity of the algorithm, to accommodate for the changing levels of insulin resistance throughout pregnancy.