Skip to main content
Premium Trial:

Request an Annual Quote

Epsilon Group Launches FDA-Approved Type 1 Diabetes Simulator


By Uduak Grace Thomas

The Epsilon Group, a Charlottesville, Va.-based medical research firm, has launched the Type 1 Diabetes Metabolic Simulator, or T1DMS, an in silico diabetes simulator based on technology licensed from the University of Virginia.

According to the company and UVA, the simulator is the first in silico tool to receive the US Food and Drug Administration's approval as a substitute for preclinical animal testing of type 1 diabetes control strategies.

Specifically, T1DMS, which is designed to simulate the glucose dynamics of an individual or population, was approved in 2008 for testing treatment strategies that involve glucose measurements and insulin injections, Marc Breton, a researcher in UVA's Center for Diabetes Technology and a co-developer of the technology, told BioInform.

The UVA team developed the simulator in collaboration with researchers from Italy's University of Padova with funding from the Juvenile Diabetes Research Foundation and the National Institute of Diabetes and Digestive and Kidney Diseases.

Epsilon licensed the underlying technology from UVA in April and launched a "training" version of the software last month. The company believes that T1DMS can not only reduce the time and cost of current preclinical animal testing methods, but can also serve as a "learning environment" for study investigators, healthcare providers, and diabetes patients.

The company plans to offer a web-based simulation service based on T1DMS in early 2012 and said it will continue to upgrade the system as new treatments come on the market.

UVA said in a statement that the tool improves on other diabetes simulators in that it allows in silico preclinical experiments "to be conducted at the level of an individual, revealing inter-personal differences due to treatment," while other programs provide only average or group-level results.

Based on a simulated cohort of 300 children, adolescents, and adults with type 1 diabetes, the T1DMS algorithm uses 26 different parameters to mimic human metabolism at the individual level, through several distinct patient profiles. Within these individual profiles, variables such as diet, exercise behavior, and insulin intake can be manipulated to test the accuracy or effectiveness of a new therapy under varying conditions, or to compare it to existing products.

The 300 patients that form the core of the simulator represent the "true" population of type 1 diabetics, Breton told BioInform. He said this in silico population was created by a mathematical model that was fitted on a set of 400 real patients using 39 metabolic parameters including body weight, blood volume, and the rate at which insulin is cleared from the system.

With this information in hand, "we created a probability distribution so we had an idea of the spread of these parameters across the population and their likelihood," Breton explained. "Based on that distribution ... we generated these 300 patients based on that initial set of 400."

The model can be used to generate other populations if necessary, he added. While the FDA-approved population is for research groups looking to move into human clinical trials, the researchers have created a second patient population, also comprised of 300 individuals, that "can be used to design the clinical trial," Breton explained. That population has not been approved by the FDA.

To use the simulator, the user selects a test scenario, such as a patient's eating and exercise habits and insulin injection schedule. Once that’s done, the user enters a possible treatment strategy into the simulator and then selects the test population for comparison.

"Once you have set these three parameters, the computer will crunch the reaction of the selected patient to the specific scenario and treatment strategy selected and compute a suite of outcome measures that have been usually associated with that type of trial — [for example] mean glucose, variability of glucose, hypoglycemia the patient suffered during the simulation, and so on," Breton explained.

"The claim is that [for] any diabetic out there, we can find a simulated patient that totally resembles his or her reaction to glucose, insulin, and meals," he said.

Breton is also involved in research projects focused on finding links between genes and patients' responses. Right now, "there is no explanation for the characteristics that these patients display and the idea would be to add a genetic component to that so that we could explain some of their characteristics," he said.

The UVA researchers will continue to play a development role in the software now that it is part of Epsilon's product portfolio. At present, Breton said, they are working on incorporating additional treatment strategies involving substances like glucagon — an injectable form of which is used in severe cases of hypoglycemia — but these have not been approved by the FDA for simulations at present.

Furthermore, the team has a JDRF grant to continue developing additional functionalities as well as to add new medical devices. These updates will also be provided for commercial users of the platform.

Currently the simulator is being used by research groups that are part of the JDRF consortium including the University of California, Santa Barbara; Stanford University; and Washington University, among others. Commercially, companies like Roche Diagnostics and Medtronics are using the platform.

Computerized Testing, Training, and Treatment

UVA granted Epsilon an exclusive license to the simulation technology in April, giving it the right to both use and commercialize the technology, Miette Michie, interim executive director and CEO of the UVA Patent Foundation, told BioInform.

She said that although several unnamed groups indicated interest in the technology, the foundation chose Epsilon because of its nearby location and also because it is a subsidiary of Medical Automation Systems, a healthcare informatics company that has in the past licensed technology from UVA.

Michie could not provide specific details about the license or any financial arrangements with the university and inventors.

Epsilon acts as a contract research organization and consulting firm to help pharma and medical device companies design and conduct clinical trials. Although the group works primarily in diabetes and associated complications, it has also worked on projects in the fields of oncology, nephrology, and geriatrics.

Epsilon expects that T1DMS will eventually replace animal tests for preclinical testing of diabetes interventions. It claims that so far, the FDA has granted four investigational device exemptions for diabetes control products based on in silico simulations on the platform.

Gail Kongable, Epsilon's vice president of research analytic services, told BioInform that the training version of the software contains 30 in silico patients, a simulated glucose sensor that measures glucose levels at regular intervals, and a simulated insulin pump that gives the in silico patients insulin injections based on their glucose levels. Users can manipulate these features to create their desired test conditions.

The web-based program that Epsilon is developing will have two service options: the FDA-approved 300-patient cohort, which can replace animal tests; and an in silico population of about 100 patients with individualized parameters that will provide a more "specific examination of the patient profile and response," she said.

The company expects its web-based service to help standardize research and development of new diabetes management tools and interventions. Through the service, the company will be able to guide users through the development process with more direct support than will be available to users who purchase the software and run it internally.

The training software is priced between $500 and $700, while the price for the service option will vary depending on the complexity of the analysis and the number of simulations required, Kongable said.

All Simulators Are Not Equal

T1DMS is not the only simulation tool available for type 1 diabetes. For example, Entelos offers the Type 1 Diabetes PhysioLab platform, an in silico model of the non-obese diabetic mouse used to study the disease that it developed in collaboration with the American Diabetes Association.

Comparing the T1DMS simulator to the Entelos platform, UVA's Breton explained that the PhysioLab simulations are based on "the average response" of the population, meaning that "there isn't a population of patients in the Entelos simulator; there is only one patient [and] there is only one reaction possible from any given input." By comparison, the UVA-developed system is able to "quantify" variability in response to specific treatments, he said.

These differences in design make both platforms appropriate for different purposes, according to Breton.

The Entelos system allows users to "run long-term simulations and try to predict the effect of drugs," he said, making it better suited for exploring new targets for drugs by shedding light on the average response of the population to specific and treatments.

However, the FDA "didn't want to only see what the average of the population would do," he said. "They also wanted to see the best and the worst case ... they want to see how dangerous a treatment is and to [do that] you have to have a notion of how variable the response to that treatment is going to be."

The fact that the UVA system offered that capability is "what convinced the FDA that [T1DMS] was capable of replacing animal trials," he said.

Entelos could not be reached for comment prior to press time.

Have topics you'd like to see covered in BioInform? Contact the editor at uthomas [at] genomeweb [.] com.

The Scan

Study Links Genetic Risk for ADHD With Alzheimer's Disease

A higher polygenic risk score for attention-deficit/hyperactivity disorder is also linked to cognitive decline and Alzheimer's disease, a new study in Molecular Psychiatry finds.

Study Offers Insights Into Role of Structural Variants in Cancer

A new study in Nature using cell lines shows that structural variants can enable oncogene activation.

Computer Model Uses Genetics, Health Data to Predict Mental Disorders

A new model in JAMA Psychiatry finds combining genetic and health record data can predict a mental disorder diagnosis before one is made clinically.

Study Tracks Off-Target Gene Edits Linked to Epigenetic Features

Using machine learning, researchers characterize in BMC Genomics the potential off-target effects of 19 computed or experimentally determined epigenetic features during CRISPR-Cas9 editing.