· Glycaemic management of hospitalised children with diabetes and complex nutritional needs is challenging. The commonly used ‘variable rate intravenous insulin infusion’ is resource-intensive and has a suboptimal safety profile. Fully closed-loop insulin delivery systems can improve glycaemic control, reduce the risk of hypoglycaemia, and significantly decrease the demand on medical and nursing resources.

· Glycaemic management of hospitalised children with diabetes and complex nutritional needs is challenging. The commonly used ‘variable rate intravenous insulin infusion’ is resource-intensive and has a suboptimal safety profile. Fully closed-loop insulin delivery systems can improve glycaemic control, reduce the risk of hypoglycaemia, and significantly decrease the demand on medical and nursing resources.

Variable rate intravenous insulin infusion (VRIII) is commonly used to address the complex nutritional needs of hospitalized patients with diabetes [1]. The process involves titration of intravenous insulin according to hourly blood glucose (BG) level measurements and is the standard of care to achieve optimal glycemic target levels in hospitalized patients with diabetes [1]. However, VRIII has inherent limitations. First, it is a reactive approach to glucose profiles rather than being based on physiology. The method does not replicate the dynamic interactions between glucose turnover rates and insulin secretion. Second, VRIII does not account for individual variations in insulin sensitivity or carbohydrate intake [1,2]. Third, the requirements for frequent monitoring of BG levels and adjustment of insulin infusion rates are extremely resource intensive. Last, patient safety is a concern with VRIII as severe hypoglycemia and other adverse events have been reported with its use [2].

Hybrid closed-loop (HCL) delivery systems are transforming the management of patients with type 1 diabetes [3]. In these systems, automated basal insulin delivery is controlled by an algorithm based on the data from real-time continuous glucose monitoring (CGM). The system does require user initiation of a mealtime bolus, and quantification of carbohydrate intake [3]. However, fully closed-loop systems (FCLs) automatically regulate insulin delivery via an insulin pump in response to CGM-derived glucose concentrations without the need for manual input of carbohydrate intake [4]. CamAPS HX is a commercially available FCL system that improves glycemic outcomes in adults with diabetes [4]. The CamAPS HX algorithm is an Android phone application (app) that uses the Cambridge adaptive model predictive control algorithm. The app receives glucose data from the Dexcom G6 sensor and adjusts the insulin delivery rate every 8–12 minutes using a Dana Diabecare RS pump (Diabecare) [5]. The system is initialized by input of patient body weight and total daily insulin dose and has a nominal glucose target level of 5.8 mmol/L. This level can be adjusted in the range of 4.4 to 10.0 mmol/L based on clinical circumstances. The system also has "boost" and "ease-off " functions that can modulate the algorithm and be used to manage hyperglycemia and hypoglycemia, respectively. The low- and high-glucose alarms of the system can be customized to suit the patient's clinical condition and individual requirements [4]. The glucose and insulin data are automatically uploaded to the Diasend (https://diasend.com//en) diabetes management platform.

In this report, we present a case of a hospitalized adolescent with diabetes secondary to acute pancreatitis whose condition was managed using a FCL system.

A 14.2-year-old previously healthy male adolescent presented to his local hospital with a 1-day history of central abdominal pain and nonbilious vomiting. On initial assessment, he was hemodynamically unstable, with tachycardia and a prolonged capillary refill time, and had abdominal tenderness with guarding. Initial laboratory results showed a high serum amylase level of 1,301 U/L (12–118 U/L) and a raised serum creatinine level of 105 umol/L (34–71 umol/L). A computerized tomographic scan showed features of acute necrotizing pancreatitis with associated splenic and portal vein thrombosis. A pneumomediastinum was also noted, and an esophageal perforation as a result of violent vomiting was suspected. He was initially stabilized with an intravenous fluid bolus. A nasogastric tube was inserted and intravenous antibiotics were administered. He was transferred to the pediatric intensive care unit (PICU) the following day due to persistent hemodynamic instability. While in the PICU, his condition was managed conservatively; enteral feeding was stopped, and intravenous fluid administration was continued. The patient did not require respiratory support or inotropic therapy.

Three days after hospital admission, the patient developed persistent hyperglycemia; his BG levels ranged between 11.8 and 15.2 mmol/L with no ketosis. The low random C-peptide level of 12 pmol/L and BG of 14.4 mmol/L suggested hyperglycemia secondary to impaired pancreatic beta-cell function. Treatment with VRIII was started on day 3 to manage hyperglycemia. This required point of care BG testing and adjustment of insulin infusion rate every 1 to 2 hours based on measured BG levels. Total parenteral nutrition (TPN) was started 6 days after admission and required revision of the VRIII insulin administration rate and initiation of subcutaneous insulin glargine (0.3 units/kg). The total daily dose of insulin ranged from 0.9 to 1.2 units/kg during administration of TPN.

The clinical course was complicated by necrosis of nearly all pancreatic tissue and by the onset of peri-pancreatic collections. CGM (Dexcom G6) was started on day 11, and VRIII rate was continually revised according to changes in composition and infusion rates of TPN, including unplanned breaks due to a blocked intravenous line. On day 12, TPN duration was reduced from 24 hours to 20 hours and included a 2-hour "wind-down" period before the break. The VRIII administration schedule was adjusted accordingly, as shown in Fig. 1.

Enteral nutrition was started gradually on day 16 via nasogastric tube, along with weaning of TPN. These concurrent activities mandated frequent revision of VRIII rate, and insulin requirements were 1.7 units/kg/day. However, tolerance to the enteral feeds was variable, with frequent vomiting and large nasogastric aspirates. His undulating clinical course with frequent changes in tolerance to nasogastric feeding and repeated revision of TPN rates led to unscheduled changes in carbohydrate intake that challenged the glycemic control. Furthermore, the peripancreatic collections persisted, and therefore he was not anticipated to establish substantial enteral nutrition for several days. The diabetes multidisciplinary team felt that neither VRIII nor conventional insulin pump therapy was unlikely to provide safe and effective glycemic management and supported the decision to use the CamAPS HX FCL system. Institutional approval for off-license use was acquired, and the system was started on day 20. A glucose target of 7 mmol/L, higher than the nominal target of 5.8 mmol/L, was set to avoid hypoglycemia. The patient, his caregivers, and the ward staff were provided training for the use of the system, responding to alerts and alarms, and use of additional functions like "boost" and "ease-off. These functions respectively increase and decrease delivery of insulin in episodes of hypo- or hyperglycemia.

The system-maintained stable glucose levels with no adverse effects during TPN administration, TPN cessation, and enteral feeding reintroduction (Fig. 2, Table 1). The CamAPS HX FCL maintained glucose levels in the target range 89% of the time, with no hypo- or hyperglycemic episodes. The time to patient recovery was shorter than anticipated; and nutrition was provided solely through enteral feeding by day 23, four days after onset of CamAPS HX system use. The patient was transitioned to the CamAPS FX (HCL system) on day 25, just prior to his discharge from the hospital.

Eighteen months after admission, insulin pump therapy was discontinued with minimal insulin requirements and an glycosylated hemoglobin level of 42 mmol/mol. He continues to receive insulin aspart injections at mealtimes. A genetic analysis performed as part of the diagnostics for acute pancreatitis revealed a pathogenic variant in the cationic trypsinogen gene (PRSS1), leading to a diagnosis of PRSS1 associated hereditary pancreatitis.

Written informed consent was obtained from patient's parent.

Evidence supporting the safety and effectiveness of FCL systems has been emerging in recent years [4]. The advantages of FCL system use for hospitalized patients include automated insulin delivery without need for an insulin bolus at mealtime, the ability to adjust insulin delivery in response to clinical conditions, and its considerably lower resource requirements like direct nursing or parental care than VRIII [6]. Also, dramatic postprandial glycemic fluctuations and increased hypoglycemia risks after consumption of large meals, potential limitations of the system, are less likely in hospitalized patients receiving limited enteral nutrition [6]. In this case report, use of the FCL system achieved optimal glycemic control with no occurrences of hypoglycemia and few hyperglycemic events. The system did not require input from staff to manage changes in enteral or parenteral modes of nutritional support, but the patient’s quick clinical recovery limited the time available for system evaluation.

The limitations of VRIIIs for glycemic control are well-documented [2], but few assessments of the effectiveness of FCL systems employing subcutaneous insulin infusions in hospitalized patients have been performed (Table 2). In 2018, Bally et al. [7] reported successful use of an FCL system based on a model predictive control algorithm in adults with type 2 diabetes receiving noncritical care. In their study, the FCL system-maintained glucose levels within target ranges 25% greater than conventional therapy. In 2019, Boughton et al. [8] reported the use of a similar system, FlorenceD2W-T2, using faster-acting insulin aspart in managing nutrition in noncritically ill adults with non-type 1 diabetes receiving insulin injections. One-third of the study patients received parenteral nutrition for a mean period of 6.7 days, and use of the FCL system markedly improved the time spent in target glucose range, 68% versus 32%, with no related adverse events. This FCL system was the predecessor of the CamAPS HX system, which showed similar efficacy in managing post-operative nutrition in adults with non-type 1 diabetes, including those who underwent partial or total pancreatectomy in clinical trials [9,10]. Furthermore, when implemented in routine clinical practice to treat hospitalized adults with non type 1 diabetes who were difficult to manage with standard insulin therapy, the CamAPS HX system was effective and safe; the system-maintained glucose levels within the target range 54% of the time with no reports of severe hypoglycemia or diabetes-related ketoacidosis [5].

The CamAPS HX system requires minimal input and is relatively simple to manage after proper training. Information on this training is available via open access [5], and a detailed training module is available online (https://hx.camdiabtraining.com/home.html). Patients experiencing significant enteral nutrition intake such as nasogastric bolus feeding may develop hyperglycemia when using this system. However, an insulin bolus can be administered to manage postmeal hyperglycemia and the system can be transitioned to a HCL system, as with our patient.

Our patient displayed insufficient insulin secretion with low serum C-peptide levels during hyperglycemic episodes, similar to type 1 diabetes. To our knowledge, this is the first report of successful in-hospital implementation of an FCL insulin delivery system to manage hyperglycemia in an acutely ill adolescent with diabetes. A previous study with a small population of 16 adolescent subjects with type 1 diabetes in a camp setting showed no difference in glycemic control between HCL and FCL systems [11]. Studies with larger samples sizes designed to evaluate the effectiveness and safety of the CamAPS HX FCL system in both adult (NCT04977908) and adolescent (NCT05653050) type 1 diabetes patients are in process. Continued evolution of these technologies will increase quality of life in young people with diabetes.