Objective
Suboptimal insulin replacement is a major cause of increased glycemic variability and worsened glycemic control in individuals with Type 1 Diabetes (T1D). Among the factors complicating insulin dosing in T1D, fluctuations of insulin sensitivity (SI) triggered by various psychobehavioral and physiological factors (e.g., circadian rhythms, physical activity, phycological stress) play a significant role. In this context, as part of a previous project sponsored by JDRF, we developed a novel method for the tracking of SI from data easily collectable and readily available to individuals with T1D (i.e., glucose sensor and insulin pump data); the tracking algorithm is able to provide a real-time estimate of a subject's SI level, which is then used within a "smart" bolus calculator to inform prandial insulin dosing. The advisory system just described was tested within a pilot clinical trial of 15 adults with T1D at the University of Virginia, where it allowed to appropriately reduce the dinner insulin dose following an early-afternoon aerobic exercise session, so mitigating the occurrence of postprandial hypoglycemia, without increasing exposure to hyperglycemia. Due to the encouraging results obtained, as part of the current project, we propose to extend the testing of the available method and investigate possible algorithmic improvements, with the ultimate goal of finalizing the design of a smart bolus calculator that can be used at home, in free-living conditions to help individuals with T1D making daily treatment decisions. In other terms, with this study, we intend to build an advisory system that is able to tailor treatment decisions to the specific metabolic state of an individual; prove the safety and feasibility of system use in semi-controlled challenging conditions (i.e., a one-week "camp" in adolescents with T1D involved in daily physical activities and exercise sessions) and unsupervised free-living scenarios (i.e., a 2-week home use of the SI-informed system in adults and adolescents with T1D); and demonstrate the system efficacy in improving glycemic control as compared to standard therapy. In addition, we will assess users’ expectations, hurdles, psychobehavioral challenges, and the resulting system acceptance while using the SI-informed bolus calculator at home, by means of standard questionnaires completed at the beginning and end of the study, to explore difficulties around the use of this technology and understand its potential psychobehavioral benefits.
Background Rationale
The quality of glycemic control in Type 1 Diabetes (T1D) is heavily dependent on multiple daily treatment decisions by the patient, to account for a wide variety of factors influencing insulin demand (e.g., circadian rhythms, physical activity, food, stress). In this context, despite the improving accuracy of glucose monitoring and the growing development of decision support systems, insulin misdosing remains common in T1D, leading to excess mortality and complication rates that are still significantly higher when compared to the general population. Insulin sensitivity (SI) is the main mediator of the effect of treatment decisions on glycemic variability and indicator of a person’s insulin needs. Despite the attempt to compensate for daily SI variations through individualized time-varying insulin therapy parameters (i.e., basal rate, insulin-to-carbohydrate ratio, and correction factor profiles), superimposed SI fluctuations happen very frequently in the life of individuals with T1D, making insulin dosing very difficult to tune. Indeed, several psychobehavioral and physiological factors influence insulin requirements. Phases of the menstrual cycle have been shown to dramatically impact SI in women with T1D, with insulin treatment decisions for females during menstruation being a guessing game. Stress has also been documented as a condition affecting SI and insulin action, and the impact of physical activity on immediate and delayed SI changes represents a highly investigated research topic. This elevated intra- and inter-day SI variability and the acknowledgement of SI as a key variable to be considered in daily treatment decisions strongly supported the development of an SI-informed bolus calculator to help patients and clinicians in the design of insulin therapy strategies that can result in optimal glycemic control. In this context, the current research project aims at finalizing and clinically validate a "smart" bolus calculator informed by SI as a way to automatically tailor daily treatment decisions to the specific metabolic state of a subject, with the ultimate goal of improving glycemic variability and quality of glycemic control in individuals with T1D.
Description of Project
Insulin dosing in Type 1 Diabetes (T1D) is oftentimes complicated by a wide variety of factors influencing insulin demand. Indeed, several psychobehavioral and physiological factors have been shown to affect insulin requirements through variations of sensitivity to insulin action (SI), such as physical activity, psychological stress, and hormonal changes across the menstrual cycle in women. Despite the attempt to compensate for SI variations through individualized, time-varying insulin therapy parameters (i.e., basal rate, insulin-to-carbohydrate ratio, and correction factor profiles), acute SI fluctuations superimposed to systematic SI changes are difficult to handle, representing a major cause of increased glucose variability and worsened glycemic control. This elevated intra- and inter-day SI variability and the acknowledgement of SI as a key metabolic parameter in T1D have recently encouraged us to investigate the possibility of informing insulin dosing with real-time SI assessments, in the attempt to tailor treatment decisions to the actual insulin needs of a person. Specifically, as part of a previous project sponsored by JDRF, we developed a novel method for the tracking of SI from readily-available glucose sensor and insulin pump data; the method is able to provide a real-time estimate of a subject's SI level, which is then used within a "smart" bolus calculator to inform prandial insulin dosing. The advisory system was tested in a pilot clinical trial of 15 adults with T1D in semi-controlled conditions at the University of Virginia (UVA), where it allowed to appropriately reduce the dinner insulin dose following an early-afternoon aerobic exercise session, so mitigating the occurrence of postprandial hypoglycemia without increasing hyperglycemia (percent time in hypoglycemia below 70 mg/dl in the postprandial period was reduced from 15% with the use of a standard bolus calculator to 8% with the SI-informed system, with almost half of hypoglycemia rescue treatments needed during the study). Due to the encouraging results obtained, we now propose to extend the testing of the available method and investigate possible algorithmic improvements, with the ultimate goal of finalizing the design of a smart bolus calculator that can be used at home, in free-living conditions to help individuals with T1D making daily treatment decisions. To achieve this aim, we defined three main phases as part of the current project: (1) in the first phase, we will extend the testing of the SI-informed bolus calculator over a week of use in adolescents with T1D, in a "camp" at UVA in semi-controlled conditions, during which study participants will be exposed to daily activities that are known to trigger SI fluctuations difficult to control (e.g., repeated physical activity and exercise); this will allow us to demonstrate the safety of the SI-informed system in challenging conditions, and the superiority with respect to standard therapy in terms of glycemic outcomes; (2) during the second phase, leveraging the knowledge acquired in the first part of the project and computer simulations, we will investigate possible algorithmic improvements, with the aim of finalizing the design of the SI-informed system; (3) in the third phase, the system in its final design will be tested during two weeks of home use in adults and adolescents with T1D; this will allow us to show that the use of the system in free-living conditions is safe, feasible, and effective in improving glycemic variability and quality of glycemic control with respect to standard therapy. In summary, this project will finalize the design of an SI-informed bolus calculator that can be safely and effectively used in both adults and adolescents with T1D in unsupervised free-living conditions, helping individuals with T1D managing their disease and potentially reducing the psychological burden associated to it.
Anticipated Outcome
Insulin dosing in Type 1 Diabetes (T1D) is complicated by a wide variety of factors influencing insulin needs. The overall aim of the current project is to develop an advisory system that is capable of tailoring insulin dosing decisions to the specific metabolic state of a subject. Specifically, we propose to finalize the design of a "smart" bolus calculator informed by insulin sensitivity (SI) that can be safely and effectively used to inform insulin treatment decisions in adults and adolescents with T1D in unsupervised free-living conditions. To achieve this aim, we defined three main phases as part of the current project: (1) in the first phase, we will extend the testing of the SI-informed bolus calculator over a week of use in adolescents with T1D, in a "camp" at UVA in semi-controlled conditions, during which study participants will be exposed to daily activities that are known to trigger SI fluctuations difficult to control (e.g., repeated physical activity and exercise); with this first part of the project, we anticipate to be able to demonstrate the safety of use of the SI-informed system in challenging conditions, and the superiority with respect to standard therapy in terms of glycemic outcomes (i.e., exposure to hypoglycemia, exposure to hyperglycemia, and glycemic variability); (2) during the second phase, leveraging the knowledge acquired in the first part of the project and computer simulations, we will investigate possible algorithmic improvements, with the aim of finalizing the design of the SI-informed system and demonstrate its stability and ease of use in semi-controlled conditions, prior to initiating home use; (3) in the third phase, the system in its final design will be tested during two weeks of home use in adults and adolescents with T1D; this will allow us to show that the use of the system in free-living conditions is safe, feasible, and effective in improving glycemic variability and quality of glycemic control with respect to standard therapy. In summary, this project will finalize the design of an SI-informed bolus calculator that can be safely and effectively used in both adults and adolescents with T1D in unsupervised free-living conditions, helping individuals with T1D managing their disease and potentially reducing the psychological burden associated to it.
Relevance to T1D
In Type 1 Diabetes (T1D), fluctuations of insulin requirements triggered by a wide variety of psychobehavioral and physiological factors (e.g., circadian rhythms, physical activity, psychological stress) typically complicate insulin dosing and lead to abnormal glucose variability and worsened glycemic control. In this context, despite the improving accuracy of glucose monitoring and the growing development of decision support systems, insulin misdosing remains common in T1D, leading to excess mortality and complication rates that are still significantly higher when compared to the general population. Glycemic variability is typically at the root of clinicians’ inability to safely achieve near-normal average glycemia in T1D, as reflected by hemoglobin A1c (HbA1c). While target HbA1c values of 7% or less result in decreased occurrence of micro- and macrovascular diabetes complications, the risk for severe hypoglycemia increases with tightening glycemic control. Consequently, individuals with T1D face a life-long optimization challenge: to reduce average glucose levels while simultaneously avoiding hypoglycemia, which has been indeed implicated as the primary barrier to optimal diabetes management. To achieve such an optimization, minimization of glycemic variability is necessary. In fact, significant reduction of average glycemia without increase of iatrogenic hypoglycemia is only possible if glycemic variability is constrained, to prevent blood glucose fluctuations from entering the low range. In addition, evidence has been collected to support the hypothesis that abnormal glycemic variability represents an independent risk factor for diabetes complications. Indeed, beyond establishing HbA1c as the gold standard for glycemic control, the Diabetes Control and Complications Trial concluded that: “HbA1c is not the most complete expression of the degree of glycemia. Other features of diabetic glucose control, which are not reflected by HbA1c, may add to, or modify the risk of complications. For example, the risk of complications may be highly dependent on the extent of postprandial glycemic excursions”. Consequently, reducing glycemic variability has become a priority in the management of diabetes, and current research is actively working on the development of decision support systems to help individuals with T1D making treatment decisions and prevent large glucose fluctuations due to insulin under/overdosing. In this context, the aim of the current research project is to finalize the design of an advisory system able to inform insulin dosing and tailor treatment decisions to the specific metabolic state of a subject. The proposed system consists of a "smart" bolus calculator informed by insulin sensitivity - a key metabolic parameter of diabetic people quantifying their insulin needs - that we propose to clinically validate in challenging semi-controlled and unsupervised free-living conditions. If successful, this project will lead to the finalization of a decision support system that will be able to optimize insulin dosing, contributing to reduce glycemic variability and improve glycemic control in people with T1D.