ľֱ

'Artificial Pancreas' System Boosts In-Hospital T2D Control

— Less time out of range, no increase in insulin or lows with fully-automated system

MedpageToday

This article is a collaboration between ľֱ and:

ORLANDO -- An automated closed-loop system for continuous glucose monitoring and insulin delivery -- a.k.a. an "artificial pancreas" -- improved glycemic control in the of hospitalized type 2 diabetes patients.

Time in the target range of 100 to 180 mg/dL was 65.8% with the system versus 41.5% with conventional subcutaneous insulin therapy (P<0.001), Roman Hovorka, PhD, of the University of Cambridge, England, and colleagues reported here at the American Diabetes Association meeting.

The approach didn't increase total daily insulin dose (44.4 vs 40.2 U, P=0.50) or hypoglycemia, with a similar duration of glucose under 54 mg/dL between the groups (P=0.80), and no cases of severe hypoglycemia or clinically-significant hyperglycemia with ketonemia.

"The advantage of a closed-loop system is the finely tuned, instantaneous glucose-responsive modulation of insulin delivery, with its continual adaptation to changing insulin needs during the day and between days," the group wrote in a paper published simultaneously in the New England Journal of Medicine.

Another advantage was no staff member input required.

"People spend a lot of time -- nurses, doctors -- trying to get the blood sugar right and it's just really hard to do in the chaotic environment where people are going to tests and not there for their meal or preparing for a test and they have to change to IV fluids, et cetera," commented John Buse, MD, PhD, of the University of North Carolina in Chapel Hill. "It's a mess to take care of diabetes in the hospital."

Furthermore, "severe hypoglycemia in the hospital can result in you not getting paid for the hospitalization at all. So there are consequences to hospitals in this regard," he told ľֱ.

Steven Russell, MD, PhD, of Massachussets General Hospital in Boston, whose group has been working on a similar "bionic pancreas," noted that the hypoglycemia seen with prior attempts to improve control has hampered progress, despite the broad consensus that in-hospital hyperglycemia among type 2 diabetes patients contributes to morbidity.

"Having an automated way to regulate glucose clearly is a good idea and what was needed was proof of principle with the current closed-loop systems in the hospitals," he said. However, "the particular system that they used is not something you would commercialize," he cautioned, noting that the glucose monitor used is no longer on the market.

The open-label trial included 136 adults with type 2 diabetes who were not critically ill but required subcutaneous insulin therapy during treatment in the general wards of two tertiary hospitals in Europe. Participants were randomized to closed-loop insulin delivery or the center's conventional subcutaneous insulin therapy strategy for up to 15 days or until discharge.

The closed-loop system used a Freestyle Navigator II continuous glucose monitor communicating via Bluetooth with a tablet connected to the Dana Diabecare R pump (on the market in Europe but not the U.S.) for rapid-acting insulin administration. It used a predictive control algorithm but no prandial insulin boluses or input of timing or carbohydrate content of meals. The usual care group had the continuous glucose monitor readings masked.

Single episodes of hypoglycemia with a capillary glucose measurement of less than 63 mg/dL, confirmed by point-of-care measurements, occurred in three closed-loop patients, compared with nine episodes occurring in eight control group patients, although none needed more than oral carbohydrates to treat it. Two closed-loop patients got supplemental insulin for sensor glucose measurements over 434 mg/dL for more than 1 hour.

Adverse events related to the interventions included skin irritation from the sensor adhesive or bruising at the cannula insertion site in three closed-loop and two control group patients. The closed-loop system had two sensor failures and one pump-check error.

Russell also cautioned that, although the trial didn't show a significant increase in duration hypoglycemia, the absolute amount of hypoglycemia was nonsignificantly higher and may have become significantly so in a larger study with a broader variety of patients.

Limitations included imbalance between groups, with greater in availability of sensor glucose measurements and a longer trial duration in the closed-loop group, perhaps due to greater burden of coexisting illnesses in that group.

Disclosures

The study was supported by Diabetes UK, the Swiss National Science Foundation, the European Foundation for the Study of Diabetes, the JDRF, the National Institute for Health Research Cambridge Biomedical Research Centre, and a Wellcome Strategic Award.

Abbott Diabetes Care discounted the continuous glucose-monitoring devices and sensors used in the trial and provided details for a communication protocol to facilitate real-time connectivity to the insulin pump.

Hovorka disclosed non-financial support from Abbott Diabetes Care and relationships with Diabetes UK, National Institute for Health Research Cambridge Biomedical Research Centre, a Wellcome Strategic Award, Eli Lilly, Novo Nordisk, BBraun, and Medtronic.

Hovorka also disclosed patents on the safety layer of the hybrid open-closed loop system used in the study, an overnight closed-loop insulin delivery with model predictive control and glucose measurement error model, a system for insulin delivery using glucose regulation and measurement error models, and a model-based closed loop system used in the study.

Russell disclosured relationships with many sensor and pump companies, although not those with devices used in the trial.

Primary Source

New England Journal of Medicine

Source Reference: Bally L, et al "Closed-Loop Insulin Delivery for Glycemic Control in Noncritical Care" N Engl J Med 2018; DOI: 10.1056/NEJMoa1805233.