Objective
The aim of this proposal is to characterize the changes in insulin requirements of individuals with Type 1 Diabetes (T1D) across pregnancy, and to develop simulation-based tools to adapt automatic insulin delivery (AID) throughout the weeks of gestation.
This objective will be achieved by addressing three specific aims:
1. Changes in insulin sensitivity during pregnancy will be modeled using field data from a cohort of pregnant individuals with T1D, and integrated into an already existing large-scale model of glucose-insulin metabolism, the UVA/Padova T1D Simulator. The novel T1D Pregnancy Simulator will support pre-clinical testing on new insulin interventions aimed at the pregnant population, in order to evaluate their feasibility and safety before human testing.
2. An algorithm will be developed to adjust the insulin-dosing parameters of an AID system. This algorithm will use recent patients’ data to build model of the patient’s metabolism. Then, this model will be used to simulate the effect of small adjustments to the patient’s therapy and choose the best one for the current gestational age.
3. The simulation-based algorithm for pregnancy-specific therapy adaptations will be tested on the novel T1D Pregnancy Simulator, and its efficacy and safety evaluated with respect to no pregnancy-specific adaptations in terms of time in pregnancy range (63-140 mg/dl) and hypoglycemic/hyperglycemic protection.
If successful, this project will i) further advance our knowledge on how glucose metabolism changes throughout pregnancies with T1D; ii) provide a simulation environment for pre-clinical trials during pregnancy; iii) generate the necessary pre-clinical data to validate an algorithm aimed at adapting therapy parameters during pregnancy.
Background Rationale
Over the last 20 years, a variety of technologies have been developed to help individuals with Type 1 Diabetes (T1D) reach their glycemic targets and improve their quality of life. First, simulators of glucose-insulin dynamics in T1D provided a testbed where new insulin therapy policies could be safely investigated. Thanks to these tools, the scientific community was able to design and collect pre-clinical evidence on Automatic Insulin Delivery (AID) systems. AID systems are now a clinical reality for the management of T1D, with established benefits to glucose regulation and quality of life.
Despite the established benefits AID technologies have brought to the diabetes community, their accessibility to pregnant people with T1D is still limited. Currently, most AID systems available in the United States have not received approval from the Food and Drug Administration for use during pregnancy, despite the fact that many individuals use AID systems “off-label” across gestation and recent clinical evidence showed benefits from their use. This is due to the higher vulnerability of pregnant individuals, added to the fact that glucose metabolism during pregnancy is not completely understood – especially as it pertains to changing insulin needs. As a result, to date, no simulation platform is available that can represent the metabolic variability of gestation and can be leveraged to generate pre-clinical data on the use of AID by pregnant individuals. Furthermore, it is still unclear how to adapt AID insulin-dosing profiles to accommodate the changing insulin requirements experienced by pregnant individuals.
We propose to leverage real data collected from pregnant individuals with T1D to model how insulin requirements change across gestation and to integrate these changes in a pregnancy-specific T1D simulator; this platform will provide an important tool for generating pre-clinical evidence on AID use during pregnancy, possibly fostering the regulatory approval of AID for use across gestation in the future. Furthermore, we introduce a framework for AID therapy adaptation during pregnancy. Several studies have already shown how field data can be used to generate individualized models of patients’ metabolism, which can be used as a testbed to simulate the effect of small therapy changes on the patient and safely assess the best therapy adjustment for the current gestational age. This simulation-based approach represent an effective solution for adapting AID to the fluctuating insulin requirements characteristic of pregnancies with T1D.
By achieving these goals, this proposal will fill two currently open gaps related to the development of technologies to improve glucose control during pregnancy in individuals with T1D, as it will:
1. Enable the development of a pregnancy-specific simulation platform capable of describing changes of insulin needs across gestation, which can support the design of technologies for glucose management during pregnancy.
2. Establish an individualized framework for AID therapy adjustments across gestation, which will enable the frequent adaptation of the AID system to the fluctuating insulin requirements experienced by pregnant individuals.
Description of Project
People with Type 1 Diabetes (T1D) are at higher risk for maternal and neonatal complications during pregnancy. Several studies have shown that high blood glucose levels during pregnancy can cause hypertension or pre-eclampsia in pregnant individuals, fetal demise or anomalies, difficult deliveries, and neonatal complications such as respiratory distress syndrome and hyperbirulinemia. However, maintaining a tight glycemic control is more challenging during pregnancy due to hormonal alterations, changing maternal behaviors, and fetal and placental glucose uptake. Our knowledge on how all these factors affect insulin requirements across gestation is still limited. As a result, less than 40% of pregnant individuals with T1D reach their glycemic targets.
Understanding how insulin requirements change during pregnancy, and how to adapt automatic insulin delivery (AID) accordingly, is therefore crucial to improve glycemic control across gestation and reduce the occurrence of maternal and neonatal complications. This proposal aims at characterizing how insulin sensitivity changes from conception to delivery, and to use this information to develop simulation tools that can support the development, validation and adaptation of AID throughout the weeks of gestation. In particular, this project is articulated in two parts.
In the first part of the project, field data will be collected from pregnant individuals with Type 1 Diabetes and used to estimate how insulin sensitivity (IS) changes across the pregnancy. This model of IS fluctuations will then be used to develop a T1D Pregnancy Simulator — a simulation platform that can reproduce the metabolism characteristic of pregnant individuals with T1D. This tool will allow researchers to preliminarily test the safety and efficacy of new therapies targeted at the pregnant population before testing them in real clinical trials.
The second part of the project will focus on designing a system that can suggest adaptations of insulin-dosing profiles (e.g., basal rate, carb ratios and correction factors) throughout the weeks of pregnancy, in order to adapt to the changing insulin requirements and improve glucose regulation. The system will use recent data collected from patients’ wearable devices to build a model of patient’s metabolism. Then, this model will be used to simulate the effect of small therapy changes, and determine the safer and most effective therapy adjustment for the patient.
If successful, this project will enhance our knowledge on how insulin requirements change during pregnancy. Furthermore, it will result in the development of simulation tools that can support the adaptation of automatic insulin delivery systems from conception to delivery – with the potential to result in better and safer care for pregnant individuals.
Anticipated Outcome
With this project, we expect to be able to characterize the changes in insulin requirements happening throughout pregnancies with by Type 1 Diabetes (T1D) from easily accessible patients’ data, like those obtainable from continuous glucose monitoring sensors and insulin infusion pumps.
We also expect to be able to exploit this knowledge to build two technologies aimed at assisting pregnant people with T1D. The first is a simulation platform of glucose-insulin dynamics in pregnant individuals with T1D, the T1D Pregnancy Simulator. We expect this simulator to be able to reproduce the glycemic variability observed in a cohort of pregnant individuals with T1D across different stages of the pregnancy.
Furthermore, we expect to develop a system for the adaptation of automatic insulin delivery across gestation. The system will be designed and validated in silico using the pregnancy-specific simulator of glucose-insulin dynamics. We expect the adaptation algorithm to be able to improve glycemic outcomes, in terms of time spent in pregnancy range (63-140) mg/dL, without increasing time spent below pregnancy range (<63mg/dL), with respect to no pregnancy adaptation. This adaptation algorithm will be tested in simulation by performing an in silico clinical trial on the T1D Pregnancy Simulator. Success will be determined by an improved time in pregnancy range and a non-increased time below pregnancy range with respect to no therapy adaptation.
Relevance to T1D
The significance of this proposal stems from the need of providing pregnant people with Type 1 Diabetes (T1D), who are at higher-risk for diabetes-related complications, ad hoc technologies to improve their glucose regulation and reduce the risk for adverse pregnancy outcomes.
The first results of this study will be the development and validation of a simulation platform reproducing the glycemic variability of pregnant individuals with T1D. This platform will represent an important tool to preliminarily test new technologies and treatment targeting this higher-risk population, and to collect further evidence on the use of automatic insulin delivery (AID) systems during pregnancy – possibly fostering their approval in the future.
Furthermore, this proposal aims at developing a framework for individualized and pregnancy-specific adaptations of AID across gestation. Currently, our knowledge on how insulin requirements change during pregnancy is still limited, and so is our knowledge on how AID should be adapted across the gestational ages, resulting in sub-optimal glycemic control. An expert system providing frequent and personalized adaptations of AID can potentially improve glycemic regulation across pregnancy and reduce the occurrence of maternal and neonatal complications.