A novel investigational fully automated “artificial pancreas” system that combines rapid-acting insulin aspart (Fiasp, Novo Nordisk) and pramlintide (Symlin, AstraZeneca) performed similarly to a Fiasp-alone hybrid system that requires carbohydrate counting in patients with type 1 diabetes, results from a small new study show.
Automated insulin delivery (AID) systems — also called closed-loop or artificial pancreas systems — combine an insulin pump, a continuous glucose monitor, and a software algorithm that allows the two devices to automate insulin delivery to achieve target blood glucose levels.
Current commercially available devices use only insulin and are hybrid closed-loop systems, meaning that they stop insulin delivery to prevent hypoglycemia and make adjustments in response to hyperglycemia, but patients must still notify the system that they are about to eat, and in some cases also enter in approximate carbohydrate counts.
Systems that combine insulin and glucagon are also currently in development. These would add extra protection against hypoglycemia, particularly during exercise, but are not proven to reduce nocturnal or postexercise hypoglycemia, and their long-term effects are unknown, said lead author of the new research Michael A. Tsoukas, MD, assistant professor, Faculty of Medicine and Health Sciences, McGill University, Montreal, Quebec.
In contrast, Tsoukas and colleagues studied a system that combines the ultra rapid-acting insulin Fiasp with the amylin analog pramlintide, which both slows gastric emptying and suppresses glucagon secretion. The aim of combining the two hormones is to minimize postmeal glucose spikes to the point where patients wouldn’t need to announce meals or count carbohydrates, an error-prone process that many patients find burdensome.
Tsoukas presented the new data during the virtual European Association for the Study of Diabetes (EASD) Annual Meeting 2020.
Aim Is to Alleviate Carb Counting When Using AID System
In the randomized crossover noninferiority trial, the Fiasp + pramlintide system (fixed ratio of 1 U Fiasp/10 µg pramlintide) produced similar blood glucose time-in-range (3.9-10.0 mmol/L or 70-180 mg/dL) without meal announcement compared with Fiasp-only with meal announcement.
Postmeal glucose levels above 10.0 mmol/L (180 mg/dL) — but not levels above 13 mmol/L (250 mg/dL) — were higher with the dual-hormone system, however. (The two hormones were delivered by separate pumps to mimic a coformulation, which is being developed.)
“We’re happy to report that the fully closed-loop, dual-hormone system, without meal announcement or carb counting, performed comparably to the hybrid closed-loop system with the exception of an increase in transient, mild hyperglycemia postprandially,” Tsoukas told delegates.
Study coauthor Ahmad Haidar, MD, also an assistant professor at McGill, told Medscape Medical News: “We were expecting to see some hyperglycemic excursions since it is a fully AID system [and] gives automatic ‘postprandial’ boluses, usually within 30-40 minutes. The aim with this system was not to further improve glucose control compared to the hybrid AID system, but to alleviate the need for carbohydrate counting while maintaining the same level of glucose control.”
Asked to comment, session moderator Mark Evans, MD, of Addenbrookes Hospital, Cambridge, UK, told Medscape Medical News that he thought the data were “fantastically exciting and will be of interest to many people.”
Evans does not see the mild postprandial glucose elevations as a problem.
“It’s normal for blood glucose to go up after meals. Sometimes we forget that. It was up compared to the announced meals, so patients spent a bit more time in ‘mild’ hyperglycemia.” That might mean a trade-off, he said, whereby patients might have a choice between a system with meal announcements in order to obtain slightly better time-in-range versus accepting slightly higher postprandial values for the freedom of not worrying about determining meal composition.
“Wouldn’t it be great if people had flexibility? If people are happy to do something in advance of a meal, that is one system. If people prefer to just stick the thing on, turn it on, and let it run without having to think about meals and snacks at all to the extent of preannouncing, that could be another system.”
“Of course the holy grail is to have a system where people really, really, don’t have to think too much about the system they’re using,” Evans commented.
Deeply Impressive Data as Two Systems Produce Similar Time-in-Range
The study involved 23 patients with type 1 diabetes randomized to 24 hours on each of the two systems with a washout period inbetween. The two groups were given identical meals, and half exercised between lunch and dinner. The patients were a mean age of 35 years, had a diabetes duration of 21 years, a mean A1c of 8.1%, and a total daily insulin dose of 0.68 U/kg.
The primary outcome, total time in target, did not differ between the two groups: 79% with Fiasp alone and carb counting versus 75% with Fiasp + pramlintide and no carb counting/meal announcement (P = .38).
Time spent above 10.00 mmol/L (180 mg/dL) was 17% with Fiasp alone versus 22% with Fiasp + pramlintide (P = .11), while the difference was far less in time spent above 13.9 mmol/L (250 mg/dL), at 0% with Fiasp alone versus 2.1% with the combination (P = .69).
Profiles were similar after each meal, with the glycemic excursions following Fiasp + pramlintide generally occurring during the first 2 hours after a meal and aligning with Fiasp alone 3 to 4 hours later, Tsoukas noted.
For time spent in hypoglycemia below 3.9 mmol/L (70 mg/dL), the combination yielded zero time, versus 2.1% for Fiasp alone (P = .068). The number of hypoglycemic events requiring treatment was nearly double for Fiasp alone (21 vs 11).
During the daytime, Fiasp alone was superior to the combination for overall time-in-range (80% vs 67%; P = .023), while the reverse was true overnight, although the result was not significant (88% vs 92%; P = .31).
And again, time spent above 10 mmol/L (180 mg/dL) was higher with the combination system during both day and night, and significantly so during the day (31% vs 19%; P = .013), but there was no difference in time above 13 mmol/L (250 mg/dL).
Gastrointestinal side effects were significantly higher with the combination than Fiasp alone (30% vs 9%). This is not surprising given that pramlintide is an enteral hormone that exerts its effects on gastric motility, Tsoukas noted. Mild gastrointestinal effects included nausea (17% vs 4%), bloating (13% vs 4%), and heartburn (4% vs 0%).
None of the Fiasp-alone participants experienced nonmild nausea, bloating, or heartburn, while these adverse effects occurred in four, one, and one patient, respectively, in the combination group.
It’s possible that these effects could be mitigated by adjusting the fixed ratio or extending the run-in or titration period, Tsoukas said, adding that the gastrointestinal side effect profile was similar to that of other enteral hormones, including the glucagon-like peptide-1 (GLP-1) receptor agonists.
Overall, Evans commented: “I was deeply impressed that the system did as well as it did. It really depends on what you really want to get out of closed-loop automation.”
Tsoukas has reported receiving grants from Eli Lilly and lecture fees from Novo Nordisk. Haidar has reported receiving research support, intellectual property purchase fees, and/or consulting fees from AgaMatrix, Dexcom, Eli Lilly, and Medtronic, and has pending patents in the artificial pancreas area. Evans has reported being an advisory board member of, speaker for, and/or grant recipient from Novo Nordisk, Dexcom, Medtronic, Abbott, Eli Lilly, and Roche.
EASD Annual Meeting. Presented September 23, 2020. Abstract 174.