Abstract
Automated insulin delivery (AID) is the gold standard of treatment for type 1 diabetes mellitus (T1DM). However, little data on the use in people on peritoneal dialysis (PD) exist. To share real-world experience on the use of AID (Medtronic MiniMed 780G SmartGuard and Guardian 4 continuous glucose monitoring system) in a woman living with type 1 diabetes on PD, we report the implementation despite the complex setting of type 1 diabetes with multiple diabetes-associated complications, PD, visual impairment, and a complete language barrier. Glycemic control 14 days before the start of AID on multiple daily injections and 9 months after the conversion is presented. AID might be a safe and promising therapeutic option for individuals with T1DM and end-stage kidney disease requiring PD. Further research is needed to demonstrate both safety and efficacy in this special population, and assessment of patient-reported outcome measures.
Introduction
In 2021, around 8.4 million people worldwide were living with type 1 diabetes mellitus (T1DM). 1 Achieving the international recommended treatment targets (hemoglobin A1c below 7% (<53 mmol/mol) or below 6.5% (<47.5 mmol/mol) without significant hypoglycemia, time in range (TIR) 70–180 mg/dL above 70%, time below range (TBR) <70 mg/dL below 4%, TBR <54 mg/dL below 1%, time above range (TAR) >180 mg/dL below 25%, TAR >250 mg/dL below 5%, and coefficient of variation (CV) below 36%) at an early stage may prevent diabetes-related complications later on.2,3 Diabetes technology, including the use of automated insulin delivery (AID) systems, has demonstrated improved glycemic control and quality of life.4,5 Transitioning from sensor-augmented pump therapy without insulin automation to systems with predictive low-glucose suspend with the ability of insulin discontinuation when hypoglycemia is predicted by an algorithm and more advanced hybrid closed-loop features, which adjust insulin delivery (basal rate adjustment and automated corrections) based on glucose levels, resulted in improved glycemic metrics. 6 However, achievement of treatment goals is still challenging and varies greatly depending on availability, access, and other barriers.7,8 For people living with T1DM already suffering from diabetes-related complications, therapeutic targets for glycemic control may differ. 2 Despite therapeutic improvement, diabetic kidney disease (DKD) is one of the most common complications of diabetes mellitus, and progression to end-stage kidney disease (ESKD) and substantially increased mortality are still present. 9 The safe use of modern technology in this population is still uncertain and impacted by several challenges, including impaired accuracy of continuous glucose monitoring (CGM) systems due to volume shifts on hemodialysis (HD) and glucose-rich solutions for peritoneal dialysis (PD).10,11
PD is a type of kidney replacement therapy (KRT) that patients can perform at home, using their own peritoneal membrane as a natural filter to remove excess fluid and waste from the bloodstream. A sterile dialysis solution is delivered into the peritoneal cavity through a catheter that is surgically placed, and processes of diffusion and osmosis help transfer waste and fluid into the dialysis solution.12,13 PD can be done manually, known as continuous ambulatory PD, which involves several fluid exchanges throughout the day. In contrast, automated peritoneal dialysis (APD) uses a machine to perform these exchanges overnight while the patient sleeps. When compared to traditional in-center HD, PD has several benefits: It allows patients to manage their treatment from home on a flexible schedule, promoting greater independence and making it easier to balance work and family life. PD is also associated with better preservation of remaining kidney function, a reduced risk of blood pressure fluctuations, and improvements in certain aspects of quality of life. 14 Furthermore, PD is typically more cost-effective than HD, providing significant savings for both patients and healthcare systems.14,15 Typically, when possible, both forms of KRT are offered to patients at our center, and younger patients are supported to get access to PD for the aforementioned benefits of this kind of KRT.
There is limited evidence on the use of CGM and AID systems in individuals living with T1DM and DKD on PD, leading to a lack of practical guidance.11,16 –19 Some data have been published on AID use in people living with type 2 diabetes mellitus (T2DM) on hemodialysis20,21 and CGM accuracy in small cohorts with T1DM and/or T2DM 22 either on peritoneal23,24 or hemodialysis.25,26 Further, some case reports were recently published, underlining the importance of this context in real-world settings.27,28 PD poses challenges for the use of diabetes technology that must be considered and have not yet been well studied: First, the exchange of glucose-containing peritoneal fluid leads to meal-independent glucose peaks. 29 Second, peritoneal bags include icodextrin, a substance that can potentially interfere with glucose measurements with glucose dehydrogenase-pyrroloquinoline-quinone or glucose dye oxidoreductase. The influence on CGM accuracy is not well established.23,30 –32 To date, no available AID system is approved for use in T1DM on PD.
Herein, we report the real-world experience of a woman living with T1DM and multiple diabetes-associated complications, visual impairment, a complete language barrier, and on PD switching from sensor-augmented multiple daily injection (MDI) therapy to AID (Medtronic 780G + Guardian 4 CGM). We used the CARE checklist when writing our report. 33
Case description
A 41-year-old woman (body mass index 30 kg/m2) living with T1DM since the age of 5 (diabetes duration 32 years) was first seen at our Outpatient Department of Endocrinology and Diabetology in April 2024 after referral from the Department of Nephrology. The patient moved to Vienna in 2022 after the outbreak of the war in Ukraine, where she was treated before. At the time of initial consultation, the following glycemic metrics (Ambulant Glucose Profile over 14 days) were documented: mean glucose 157 mg/dL, TBR 1% and 0% (TBR <70 mg/dL and TBR <54 mg/dL), TIR 70–180 mg/dL 70%, TAR 29% (TAR >180 mg/dL and TAR >250 mg/dL), 38.7% CV, 7.1% glucose management indicator, 7.7% (61 mmol/mol) HbA1c. The patient was on sensor-augmented MDI therapy with insulin glargine U100 and insulin aspart (total daily insulin dose 16 international units) with lots of effort during the night, when the peritoneal fluid was replaced, leading to nocturnal glucose peaks. In a peritoneal equilibration test, after the start of PD, the patient showed a high transport capacity. She experienced significant distress because of frequent blood glucose fluctuations (up to 350 mg/dL) and the need for repeated correction boluses during the night. Hence, the woman expressed the desire for less burden in managing her glucose levels, especially during sleep. She already had manifest diabetes-related complications, including DKD, diabetic polyneuropathy, peripheral arterial occlusive disease with amputations of two toes on the left and three fingers on the right, and proliferative diabetic retinopathy with bilateral visual impairment (unable to read at distance, able to recognize patterns at close range). The patient had a history of severe hypoglycemia with a seizure in 2022. A simultaneous pancreas-kidney transplantation was considered for which stable and optimized glycemic control was meaningful. Her native language was Ukrainian, and therefore, the appointments took place with an interpreter to overcome a complete language barrier (the patient did not speak German or English). Although the patient had an acceptable glycemic control on MDI, taking into account the manifest complications, the individual treatment goals addressed avoidance of hypoglycemia, improvement in CV, and an improvement in quality of life by minimizing diabetes burden, especially during the night.
Results, intervention, and follow-up
The woman was informed in detail about all available AID systems in Austria. Because of the existing language barrier (Russian/Ukrainian) and visual impairment, a system with little user-dependent input seemed suitable, which led to the decision in favor of the Medtronic MiniMed 780G and Guardian 4 with SmartGuard algorithm. The patient was admitted to the hospital for a total of 7 days to start AID therapy, because of the challenges described above and individual preferences. The schedule of the inpatient stay and follow-up visits is shown in Table 1. With the primary goal of hypoglycemia avoidance, we started auto-mode with a glucose target of 120 mg/dL and an active insulin time of 3 h. After 5 days with stable glycemic control in the inpatient setting, the auto-mode was started, after adjustments in basal rate and insulin to carbohydrate-ratio (ICR). To intercept overnight glucose peaks, the maximum basal rate was increased, the ICR was not aggressively reduced (avoidance of too early Safe Meal Bolus), and the patient was instructed to add carbohydrates before and during the ongoing dialysis. At the time of the initiation of AID, the patient was on an APD schedule, with an additional manual exchange (APD plus). The patient was advised to perform the manual exchange with the Fresenius balance with 1.5% glucose daily at 05:00 in the afternoon. At 09:00 p.m., the actual APD started and was performed with a total of six cycles, three cycles of Fresenius balance with 4.25% glucose for 25 min, and three cycles of Fresenius balance with 1.5% glucose for 50 min in an alternate fashion. During the day, an icodextrin solution was applied. The PD regimen was not changed during the AID start or observation period. The PD solution consisted of a total of 316.5 g of carbohydrates over 8 h. The patient was advised to announce 20 g of carbohydrates hourly for the first 3 h and 30 g of carbohydrates for the following 5 h overnight to increase the algorithm’s aggressiveness for the achievement of glucose targets (input of so-called “fake carbs” for meal-independent glucose peaks). Because the PD cycles mainly run during the night, we adapted the scheme with more grams of carbohydrates until the time of falling asleep, and the rest of the night was automatically regulated by auto-corrections and basal rate adjustment of the algorithm. This enabled us to achieve very stable nocturnal patterns without hypoglycemia. Throughout the inpatient stay and follow-up period of 9 months on AID, no serious adverse events occurred.
Schedule of inpatient stay, follow-up visits, and changes in SmartGuard™ setting.
During the inpatient stay, training by diabetes educators and the medical team was guaranteed as necessary.
ICR, insulin-carb-ratio.
Results
Glycemic results before and after conversion to AID are presented in Table 2 and Figures 1 and 2 for 14-day periods. TIR increased from 65% to 78% to 84%, whereas a decrease in CV 32.6% to 25.4% to 24.9%, mean glucose 157 to 150 to 144 mg/dL and TAR >180 mg/dL 33% to 22% to 16% was documented (sensor-augmented MDI 14 days before AID initiation, AID after 3 and 9 months for 14-day periods) exceeding international treatment goals. 3 The ICR was reduced over time, which increased the aggressiveness of the algorithm. Basal-bolus proportion was quite balanced, and the amount of autocorrection increased over time (Table 2). Although there was no quantitative evaluation of diabetes distress and quality of life, the patient-reported subjective relief in everyday life during the follow-up visits and enjoyed excellent quality of life compared to MDI therapy during PD. There were no changes in the diabetes-related complications, especially no progression of retinopathy.
Glycemic metrics before and after AID initiation (14-day periods).
Only descriptive data due to small sample size.
AID, automated insulin delivery; CV, coefficient of variation; GMI, glucose management indicator; ICR, insulin-carb-ratio; MDI, multiple daily injections; n.a., not applicable; SD, standard deviation; TAR, time above range; TBR, time below range; TDD, total daily dose; TIR, time in range.
Report of sensor glucose with percentile distribution over 14 days before AID start.
Report of daily sensor glucose with percentile distribution over 14 days.
Discussion
Herein, we present the case of a woman living with T1DM and manifest diabetes-related complications with the need for PD starting on AID therapy. Although the patient showed an overall acceptable glycemic control on MDI therapy, considering the severity and number of complications related to T1DM, she experienced significant diabetes distress due to frequent overnight meal-independent blood glucose peaks in the context of PD. Therefore, she was started on AID therapy to improve both glycemic control and quality of life. Our case provides new value showing a combination of complex barriers to diabetes technology implementation: The patient has severely impaired vision, uses PD for ESKD treatment, and there is a relevant language barrier.
There is very limited published data on AID use in KRT, and only one case report exists on AID in PD treatment: In the case report by Pintaudi et al. 34 advanced technology (advanced hybrid closed-loop MiniMed 780G system and real-time CGM) use in a deaf-mute patient was shown to be safely working. In our case, the impaired vision due to retinopathy was difficult, but still allowed AID implementation employing specific education and step-wise learning of skills. Relating the use of diabetes technology in the context of ESKD with KRT, one case report by Rossi et al. 28 demonstrated safe implementation of a hybrid closed-loop system with basal insulin adjustment, but without automated corrections (Medtronic MiniMed 670G and Guardian 3 CGM) in a man on PD. A case series by Chaudhry et al., 27 including four patients on hemodialysis with the use of advanced hybrid closed-loop systems CamAPS FX with mylife Ypsopump in one and MiniMed 780G and Guardian 4 CGM in three patients, also showed safe and efficient use of two different AID systems in the real-world setting. These findings corroborate the herein presented case, demonstrating the safe use with substantial improvement in glycemic metrics exceeding international treatment goals in a woman living with T1DM and on PD. Although specific AID-systems features might be beneficial for the use in people living with T1DM on KRT (e.g., Ease-off-mode 35 or Temp Target 36 ) because of limited case numbers, no comparative conclusions can currently be drawn.
To overcome visual and language barriers, we consider structured intensive training with interpreter support in an inpatient setting essential to ensure optimal safety. The optional setting of other languages in the Medtronic MiniMed 780G was also beneficial in the herein presented case. While these concepts may be transferable across various healthcare environments, their successful implementation relies on the collaboration of multidisciplinary teams and the availability of institutional resources. Therefore, the generalizability of these practices is inherently linked to the underlying healthcare infrastructure.
Some studies support the use of CGM and AID in people living with diabetes and KRT; still, further research is needed to demonstrate CGM accuracy and validate AID algorithms while the use of both hemo- and PD.16 –18 Artificial intelligence could help improve personalized algorithms to better identify patterns in glucose sensitivity and insulin clearance for patients with ESKD, instead of excluding this population from labeled use. Although glycemic targets are less stringent for people living with T1DM and advanced diabetes-associated complications, it remains important to avoid hypoglycemia, and the use of modern diabetes technology has successfully demonstrated this in the past. Furthermore, novel diabetes technologies might help to decrease diabetes distress and thus considerably improve quality of life in this population. Despite the single-patient design with limited generalizability beyond the specific clinical context described and a lack of a controlled comparison, our report supports feasible AID implementation in a high-risk setting. The outcome data is descriptive in nature, limiting the interpretability of the results. Future studies should address the cost-effectiveness, potentially improved mental well-being of individuals living with diabetes, and we emphasize the need for larger observational and well-designed prospective studies to validate and expand upon the present findings.
To our best knowledge, this represents one of the first real-world experiences of a woman living with T1DM, with advanced diabetes-related complications on PD, with significant visual impairment, a complete language barrier, and therefore pronounced barriers of technology use, who was started on AID therapy with advanced hybrid-closed-loop-therapy demonstrating highly promising results in safe and efficient use. In conclusion, our case shows the feasibility of AID implementation in a high-risk setting and highlights the need for further research.
