Artificial pancreas for pregnant diabetics

“An artificial pancreas given to pregnant women with diabetes could save mothers' lives and improve the health of their babies,” BBC News has reported. The broadcaster said the device can keep sugar at normal levels for pregnant women with type 1 diabetes, for whom blood sugar control is quite difficult.

The device, which includes a small blood glucose sensor worn on the skin, was tested in a small study of 10 pregnant women with type 1 diabetes. The monitor’s ability to measure blood sugar and adjust insulin dosage accordingly was found to have a number of benefits for the women, with glucose levels being generally well-controlled.

However, this study did not compare this approach with other forms of intensive sugar control, such as manual blood sugar testing and insulin injections. Therefore the study’s results should be regarded as preliminary until further research directly compares the device against different methods. The researchers also said that, to ensure the best outcomes for the mother and baby, the mother’s glucose may need finer regulation than seen in this study.

Where did the story come from?

The study was carried out by researchers from the University of Cambridge, Cambridge University Hospitals NHS Foundation Trust, the Diabetes Centre at Ipswich Hospital NHS Trust and the Norfolk and Norwich University Hospital NHS Trust. It was funded by Diabetes UK, the National Institute for Health Research, the Juvenile Diabetes Research Foundation, Abbott Diabetes Care, the Medical Research Council, the Centre for Obesity and Related Metabolic Diseases, Cambridge Biomedical Research Centre and Addenbrooke’s Wellcome Trust Clinical Research Facility. The research was published in the peer-reviewed medical journal, Diabetes Care.

BBC News has reported this research and its context well, providing a balanced view of the treatment’s potential future. Describing the device as an “artificial pancreas” may incorrectly suggest that it is an implantable synthetic or mechanical organ. In fact, this study was the first step in developing a system of continuous monitoring and dosing using a sensor that is taped to the arm or abdomen with a special adhesive, using a 5mm-long filament inserted under the skin to measure the level of glucose in the underlying tissue. The glucose readings from this sensor are then transmitted to a wireless receiver that can track blood sugar, and potentially control an automatic insulin delivery system that can administer adjusted insulin doses.

What kind of research was this?

This was a small observational study, without a comparison group, evaluating the effects of a technique known as “closed-loop insulin delivery” as a means of controlling blood sugar in pregnant women with type 1 diabetes. Type 1 diabetes occurs following the destruction of insulin-producing cells in the pancreas. This subsequently means the body is left without insulin, and is therefore unable to regulate the levels of glucose in the blood. The disease must be treated indefinitely with insulin, with a pancreatic transplant required in some extreme cases.

Pregnant women with type 1 diabetes find it particularly difficult to regulate their blood glucose because of pregnancy-related hormonal changes that affect the way insulin is metabolised, as well as changes in the baby’s weight and sugar requirements. Poor insulin control can lead to high sugar levels (hyperglycaemia), which in turn can lead to problems for the mother and baby.

The researchers were investigating the use of closed-loop insulin delivery for pregnant women both in the early and late stages of their pregnancy. This system continuously monitors the blood glucose of the patient and delivers insulin at the right dose when needed. The system has three important components, and this study was investigating the appropriateness of a commercially available device (called the FreeStyle Navigator) for the first two of these:

  • a way to continuously monitor the glucose levels
  • an algorithm which can be used to convert the glucose reading into an appropriate insulin dose for delivery to the patient (this is called a model predictive algorithm)
  • an insulin pump that can deliver the insulin

The women were connected to an insulin pump in this study, but dosing was not automatic as the purpose of this research was to validate the algorithm that would determine the appropriate amount of insulin. Instead, a nurse adjusted the insulin dose every 15 minutes using readings from the continuous monitoring and the algorithm.

What did the research involve?

Ten pregnant women, with an average age of 31 and with type 1 diabetes, were recruited to the study through three antenatal diabetes clinics in the UK. They were admitted to the research facility for 24-hour stays on two occasions; once early during their pregnancy (12 to 16 weeks) and again during later pregnancy (28 to 32 weeks). They were all receiving intensive insulin therapy either through the use of a pump or by repeated daily injections. All had a healthy pregnancy and in those with significant obesity, poor blood sugar control or other problems were not included.

The day before being admitted, the women had a FreeStyle Navigator sensor inserted into their upper arm and went through the device’s standard 10-hour calibration process to adjust it to their blood glucose levels. The women were then admitted to the research facility and had an insulin pump fitted to them. They were assessed following a standard evening meal and again after eating breakfast the next morning.

The researchers used the women’s weight, basic insulin requirements and total insulin dose in the preceding three days to adjust the algorithm to calculate how much insulin was required in relation to their blood glucose levels. At each session, the researchers determined the blood glucose levels and how much time the women had spent in their target glucose ranges. The researchers recorded any episodes of high blood sugar (hyperglycaemia) or low blood sugar (hypoglycaemia). They assessed overnight glucose control and glucose control around meal times (by measuring prandial insulin levels). They also determined how accurate the FreeStyle Navigator sensor was at detecting blood glucose by comparing it to independent measures of plasma glucose.

What were the basic results?

When assessing overnight glucose control, women in early pregnancy spent 84% of their time in the target range of blood glucose, and women in late pregnancy scored 100%. The women were hyperglycaemic for 7% of the night in early pregnancy but not at all during late pregnancy. No women were hypoglycaemic during the night in this study.

Around mealtimes, results were similar between early and late pregnancy, with women spending 68% to 77% of their time within the appropriate blood glucose target ranges after a large evening meal. Glucose control after the breakfast meal was achieved less well, with more women outside their target ranges compared with after the evening meals.

The FreeStyle Navigator sensor performed with no episodes of unsafe control and was deemed to be clinically acceptable about 94% of the time. There were no episodes of low blood sugar (hypoglycaemia) with symptoms. There was one unexplained episode of a woman in early pregnancy experiencing hypoglycaemia without symptoms during the early hours of the morning.

How did the researchers interpret the results?

The researchers conclude that they have demonstrated the acceptability of the FreeStyle Navigator monitoring and algorithm system in women with type 1 diabetes during pregnancy. They said that the use of this system was associated with almost-normal blood glucose overnight in both early and late pregnancy, and that this indicates that the algorithm can adjust the need for insulin as necessary over the course of pregnancy.


This small “proof of concept study” has found that a system of continuous blood sugar monitoring and automatic dose calculation appears effective and safe for women with type 1 diabetes, both early and late in their gestation. The researchers found that while using the device, none of the women had symptoms of hypoglycaemia (low blood sugar) at night. The researchers compare the results of their small study to other findings that suggest that pregnant women with type 1 diabetes spend on average 16.2% (about 1.3 hours) of the night in a state of hypoglycaemia.

The study authors also said that their system reduced the time women were hyperglycaemic (had high blood sugar) in the night. Their study found that women had blood sugar over the ideal limit 7% of the time, compared with about 36% seen in other studies.

It is important to note this is not a full product that includes continuous monitoring and automatic dosing in one device. A nurse was involved in delivering the insulin according to the continuous readings fed into the algorithm every 15 minutes. It is premature to refer to the this as an artificial pancreas as it doesn’t replace its function.

The researchers said that on the basis of these findings they are planning a randomised controlled study of closed-loop insulin delivery with tighter blood glucose targets, alongside a comparison group that will be treated with other intensive control methods. This will first take place in a hospital setting and will then be extended into the home environment. In the meantime, they said that the results from this study pave the way for future research to refine the system in pregnancy.

This is well-conducted research in an important area of medicine but it is still a small, preliminary study and the results will need to be replicated in larger studies that further explore the safety and feasibility of this system for pregnant women with type 1 diabetes. Ultimately, the purpose is to reduce death and miscarriage rates in diabetic mothers and their babies, and larger long-term studies will need to assess whether this approach to glucose control can consistently deliver such benefits: better glucose control and fewer adverse outcomes.

NHS Attribution