Status and phase
Conditions
Treatments
Study type
Funder types
Identifiers
About
Achieving near-normoglycemia has been established as the main objective for most patients with type 1 diabetes (T1DM). However, insulin dosing is an empirical process and its success is highly dependent on the patients' and physicians' skills, either with multiple daily injections (MDI) or with continuous subcutaneous insulin infusion (CSII, the gold standard of insulin treatment).
Postprandial glucose control is one of the most challenging issues in the everyday diabetes care. Indeed, postprandial glucose excursions are the major contributors to plasma glucose (PG) variability of subjects with (T1DM) and the poor reproducibility of postprandial glucose response is burdensome for both patients and healthcare professionals.
During the past 10-15 years, there has been an exponentially increasing intrusion of technology into diabetes care with the expectation of making life easier for patients with diabetes. Some tools have been developed to aid patients in the prandial bolus decision-making process, i.e. "bolus advisors", which have been implemented in insulin pumps and more recently in the newest generations of glucometers. Currently, the availability of continuous glucose monitoring (CGM) has opened new scenarios for improving glycemic control and increasing understanding of post-prandial glycemic response in patients with diabetes.
Results from clinical studies suggest that sensor-augmented pumps (SAP)may be effective in improving metabolic control, especially when included as part of structured educational programs resulting in patients' empowerment. Similarly, preliminary results from pilot studies indicate that automated glycemic control, especially during nighttime,based on information from CGM is feasible. However, automatic management of meal bolus is currently one of the main challenges found in clinical validations of the few existing prototypes of an artificial pancreas. Indeed, fully closed-loop systems where information about meals size and timing is not given to the system have shown poor performance, with postprandial glucose higher and post meal nadir glucose lower than desired. This has promoted other less-ambitious approaches, where prandial insulin is administered following meal announcement (semi closed-loop). However, despite the use of meal announcement, currently used algorithms for glucose control (the so-called PID and MPC), show results that are not yet satisfactory due to the risk of producing hypoglycemia.
One of the limitations of the current open-loop (bolus advisors) and closed-loop control strategies is that glycemic variability is not taken into account. As an example, settings of CSII consider inter-individual variation of the parameters (insulin/carbohydrates ratio, correction dose, etc.) but disregard the day-to-day intra-individual variability of postprandial glucose response. Availability of massive amount of information from CGM, together with mathematic tools, may allow for the characterization of the individual variability and the development of strategies to cope with the uncertainty of the glycemic response to a meal.
In this project, a rigorous clinical testing of a CGM-based, user-independent algorithm for prandial insulin administration will be carried out in type 1 diabetic patients treated with insulin CSII.
First of all, an individual patient's model characterizing a 5-hour postprandial period will be obtained from a 6-day CGM period. The model will account for a 20% uncertainty in insulin sensitivity and 10% variability in the estimation of the ingested carbohydrates. Based on this model (derived from CGM), a mealtime insulin dose will be calculated (referred as iBolus). Then, the same subjects will undergo standardized meal test studies comparing the administration of a traditional bolus (tBolus, based on insulin to CHO ratio, correction factor, etc.) with the CGM-based prandial insulin delivery (iBolus).
Significant advances in postprandial control are expected. Should its efficiency be demonstrated clinically, the method could be incorporated in advanced sensor augmented pumps as well as feedforward action in closed-loop control algorithms for the artificial pancreas, in future work.
Full description
Over the last 30 years, even with the development of new glucose monitoring techniques and the availability of new insulin preparations with more physiological profiles, SC continuous administration systems were still not able to be universal, efficient and safe systems able to achieve a near-normalization of glucose levels in diabetic patients. Indeed, in developed countries, only one third of diabetic patients meet criteria for good metabolic control, i.e. glycosylated haemoglobin < 7%.
During the past 10-15 years, there has been an exponentially increasing intrusion of technology into diabetes care with the expectation of improving metabolic control and making life easier for patients with diabetes. In the last years, some tools have been developed to aid patients in the prandial bolus decision-making process as the "bolus advisors", which are implemented in insulin pumps and more recently in newest generations of glucometers. Currently, the availability of continuous glucose monitoring (CGM) has opened two scenarios:
The artificial pancreas would represent the ideal solution for the attainment of the therapeutic goals needed to prevent chronic complications of diabetes. Indeed, in the last two decades, technological progresses have fuelled research on closed-loop glucose control systems aiming for effective treatment of diabetic subjects. Preliminary studies using off-the-shelf insulin pumps and continuous glucose monitoring (CGM) sensors have suggested that in research settings, closed-loop systems that automatically dispense insulin can achieve better glucose control than open-loop systems in which people have to take dosing decisions. Such promising results prompted the Juvenile Diabetes Research Foundation (JDRF) to push the research forward by launching its Artificial Pancreas Project in 2006. Also, the US Food and Drug Administration (FDA) designated the artificial pancreas as a priority within its Critical Path Initiative. However, due to its complexity, only a few prototypes so far have been developed and tested in controlled clinical settings.
Among problems related to glycemic closed-loop control, management of postprandial glycaemic excursions is a key issue in the future artificial pancreas. Indeed, meal-induced perturbations on glucose control is one of the major problems to counteract and the main challenge found in current clinical validations of the few existing prototypes of closed-loop glycemic control systems.
The first significant clinical result regarding fully automated closed loop in the fasting condition comes from Medtronic Inc. who demonstrated the feasibility of a fully automated closed loop system in 10 adults with type 1 diabetes mellitus, using an external pump (CSII), a sensor for continuous subcutaneous glucose monitoring (CGM), and a control algorithm called ePID. This algorithm consists of a classical Proportional-Integral-Derivative controller plus insulin on-board feedback. Since then, several initial clinical trials of closed-loop control have been made to prove the feasibility of other control algorithms like Model Predictive Control (MPC). MPC has obtained positive results in type 1 diabetic patients and also in Intensive Care Units.
Different approaches have been suggested to deal with meal disturbances in these controllers. Fully closed-loop systems where information about meals size and timing is not given to the system have shown poor performance, with postprandial glucose higher and post-meal nadir glucose lower than desired. This has promoted other less-ambitious approaches, where meals are announced to the system generating a feed-forward action like for instance a prandial insulin bolus (semi closed-loop). Hybrid approaches have also been proposed, where only a percentage of the prandial bolus is applied ('priming bolus') and the rest is left to the closed-loop controller.
Clinical studies have demonstrated the efficacy of these solutions to reduce postprandial excursions during closed-loop control versus fully closed-loop systems, showing that first generations of an artificial pancreas will require announcement of meals and priming insulin boluses.
However, despite the use of meal announcement, the main challenge of control algorithms is still the avoidance of overcorrection. An aggressive-enough tuning for a low post-prandial glucose peak may cause an accumulation of insulin producing a late hypoglycemia. This imposes the consideration of constraints on residual insulin activity (insulin-on-board) both in PID and MPC-based systems. However, despite the inclusion of constraints, clinical results during a meal of PID and MPC are not yet satisfactory.
Interval techniques have shown to be particularly suitable to deal with constraints under uncertainty, leading to more robust solutions and potentially reducing the risk of hypoglycaemia while maintaining good performance. These techniques were first introduced by Bondia et al in 2009, who proposed a set-inversion-based algorithm for calculation of meal-related insulin. This algorithm computed the feasible set of insulin profiles to fulfill the given constraints on postprandial glycemia, according to a patient's prediction model. In particular, physiological constraints were applied using postmeal guidelines from the International Diabetes Federation aiming at no hypoglycemia and two-hour glucose below 140 mg/dL, in a 5-hour time horizon. A refined algorithm was presented by Revert et al in 2009, allowing for the determination of the optimal insulin administration mode (standard, square, dual-wave or temporal basal decrement/iBolus). In this work, an in silico validation using the FDA-accepted UVA simulator for the testing of control algorithms was performed. Results of this study demonstrated the effectiveness of this strategy, including the challenge of meals with high carbohydrate content.
To date, priming prandial boluses in the context of semi-automated glucose control are computed based on the patient's insulin-to-carbohydrate ratio, as currently done in 'standard' CSII therapy. In this latter, bolus insulin is infused over the patient's basal insulin rate, usually following one of three available choices: 1) simple bolus (all of the insulin dose is administered as a bolus, i.e. like with a pen or syringe); 2) dual wave bolus (a percentage of the insulin dose is administered as a bolus, being the remaining insulin being infused as a square wave during a pre-specified time interval following the meal); 3) square wave bolus (all the insulin dose is administered as a square wave). However, the above mentioned study by Revert et al. has demonstrated 'in silico' (i.e. by means of an FDA-accepted computer simulator), that a coordinate action of basal and bolus insulin is required to maintain blood glucose in a physiological range, in the postprandial state. In particular, a bolus greater than the standard one, paralleled by a temporary reduction of the basal insulin infusion rate (referred as iBolus, which may be considered as a generalization of the superbolus concept introduced by Walsh et al. is needed, especially for meals with higher carbohydrate content.
This study was planned to validate this new methodology for prandial insulin administration, and it is expected to confirm the hypothesis that set-inversion techniques may be applied to SAP-CSII therapy. Of note, this strategy would represent the first attempt of developing a non-heuristic tool for mealtime insulin dosing. It could be implemented not only in closed-loop strategies of glycemic control but also in open-loop strategies as an advanced bolus advisor in newest generations of insulin pumps.
Primary objective:
In type 1 DM subjects treated with CSII, assessment and clinical validation of a new algorithm for optimization of postprandial glucose control, the iBolus (CGM-based prandial insulin administration) in comparison with a standard bolus (tBolus).
Enrollment
Sex
Ages
Volunteers
Inclusion criteria
Exclusion criteria
Primary purpose
Allocation
Interventional model
Masking
12 participants in 2 patient groups
Loading...
Data sourced from clinicaltrials.gov
Clinical trials
Research sites
Resources
Legal