'Artificial Pancreas' Is Far From Perfect, but Holds Promise

Miriam E Tucker

February 28, 2018

The MiniMed 670G hybrid closed-loop insulin delivery system (Medtronic) offers the potential for improved blood glucose control for people with type 1 diabetes, but clinicians should anticipate the need for close monitoring, frequent adjustments, and ongoing patient and family education, particularly in the first month. 

The new findings, from 31 patients with type 1 diabetes aged 14 to 26 years who wore the 670G for 3 months as outpatients, were published online February 14 in Diabetes Care by Laurel H Messer, RN, MPH, CDE, of the Barbara Davis Center for Diabetes, Aurora, Colorado, and colleagues.

The 31 "emerging adult" study participants were among 124 patients aged 14 to 75 years in the pivotal trial that led to approval of the 670G system by the US Food and Drug Administration (FDA) in September 2016.

In this deeper dive into transitioning young patients to the 670G from stand-alone insulin pumps, with or without continuous glucose monitoring, adjustments in carbohydrate-to-insulin ratio and active insulin time settings were frequently required.

In addition, "auto mode exits", whereby the system switches back into open-loop mode for safety reasons, were common in the beginning but diminished as patients more closely adhered to system instructions, including bolusing before meals. 

"The 670G is like a 1990s cell phone. It's not an iPhone 6.0. It's a giant leap forward in automation of insulin delivery, but it's not a plug-and-play system," study coauthor Gregory P Forlenza, MD, assistant professor of pediatrics at the Barbara Davis Center, told Medscape Medical News.

The authors chose to focus on adolescents and young adults for the current analysis, as these populations tend to be among the most challenging in terms of glycemic variability and stand to gain the greatest long-term benefit from automated insulin delivery.

However, the lessons learned from using the 670G system are relevant to all users, Messer and colleagues note in their article.

"There is a rather steep learning curve for both the provider and patient. If you don't anticipate that, you'll say this doesn't work. But if you anticipate and address that, it works beautifully," Forlenza said.

With Adjustments, Time in Range Improved

In the substudy group of emerging adults, participants were a mean age 17.8 years, and had a mean duration of type 1 diabetes of 9.3 years and mean baseline HbA1c of 7.8%. By the end of the 3-month study, HbA1c dropped by 0.75 percentage points (P < .0001).  

Over the same period, the carbohydrate-to-insulin ratios used to determine meal boluses, which users must still direct the system to initiate, needed to be strengthened by about 10% to 20%. This was necessary because the 670G system — unlike stand-alone insulin pumps — dials back basal insulin delivery following a meal bolus. (Traditional pumps only consider the amount of insulin "on board" when calculating correction doses.) 

Because of that, the ratios needed to be significantly decreased for more insulin to be delivered. At breakfast, the carbohydrate-to-insulin ratios decreased from 7.9:1 at baseline to 7.4:1 at week one, and leveled off at 6.9:1 for months 2 and 3 (P ≤ .001 for all time points from baseline). Similar adjustments were required for lunch and dinner. 

"Most patients won't get any auto-basal after a meal bolus. That's part of the reason we had to make the ratio more aggressive. Providers and patients need to know that's the case," Forlenza explained.

With those adjustments, the percentage of time in the target glucose range of 70 to 180 mg/dL significantly increased with auto mode use, from 55.3% at baseline to 68.4% within the first 7 days (P < .001), and the 15% improvement was sustained through the remainder of the trial.

"Small variations are what it's very good at catching up on...If you underestimated meal carbs by 10%, the [system] will minimize that, and blood glucose levels won't be as high for as long as with an open-loop. But you can't skip a meal bolus," he noted.

By the same token, the system will decrease basal insulin to prevent hypoglycemia if too much insulin is given (10% to 20%), he added. 

Still a "First-Generation System," but Improvements Being Made 

One of the challenges — which may disappear in future iterations of the hybrid closed-loop system — is the "auto mode exit," which can be performed intentionally by the user or automatically by the 670G under certain circumstances.

Prolonged hyperglycemia (> 300 mg/dL for 1 hour or > 250 mg/dL for 3 hours), which can be caused by a pump or infusion-set malfunction, will automatically cause the system to switch back to open-loop (manual) mode.    

Other situations can trigger a "safe basal" mode, in which the system calculates basal delivery but doesn't modulate it based on sensor readings.

This can happen when the system's internal checks detect a problem with the sensor, which can often be resolved by entering an extra fingerstick blood glucose for calibration (in addition to the usual two per day). If the user doesn't enter that value, the system reverts to manual mode after a 90-minute "safe basal time-out."

Auto mode was active for 87% of the time during the first 7 days of 670G use, dropping to 80% at the end of the first month, and 72% by 3 months. Most of the system-initiated "auto mode exits" were because of "safe basal time-outs."

"It's very nuanced...As this moves into the real world, we need to manage expectations," Forlenza said.

But changes are happening. On February 26, 2018, Medtronic received FDA approval for use of the Guardian Sensor 3 on the upper arm. The move is expected to increase both convenience and accuracy, according to a company statement.

Overall, Forlenza said, "We've shown that patients are doing better. They're improving their time in range by about 15% from before. That's huge. That's about 3 hours a day more time that people are in target range."

"They still have to work, and they still have high and low blood sugars, but this improves it. And once you learn the system, it improves it with less work. It's a first-generation system and there's a lot of learning required for both providers and patients. But once you get over that, people generally find it's worth it."  

The study was funded by Medtronic MiniMed. Forlenza reported receiving research grants from Medtronic MiniMed, Insulet Corporation, Tandem Diabetes Care, and Dexcom, is a consultant for Abbott Diabetes Care, and advisory board member for Dexcom. Messer has reported no relevant financial relationships.

Diabetes Care. Published online February 14, 2018. Abstract

For more diabetes and endocrinology news, follow us on Twitter and on Facebook.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.