Fast-Acting Insulin Aspart Use with the MiniMed TM 670G System

Abstract

Background:

This study assessed the efficacy and safety of ultrarapid insulin Fiasp^® in the hybrid closed-loop MiniMed™ 670G system.

Methods:

This was a pilot randomized double-blinded crossover study among established MiniMed™ 670G users comparing percentage time in range (TIR) and hypoglycemia for Novolog^® and Fiasp. After 2 weeks optimization with their home insulin, participants were randomized to receive Novolog or Fiasp for 2 weeks, followed by the other insulin for the next 2 weeks. Data from the second week of blinded insulin use were analyzed to allow 1 week for 670G adaptation. During the second week, individuals were asked to eat the same breakfast for 3 days to assess differences in meal pharmacodynamics.

Results:

Nineteen adults were recruited with mean age of 40 ± 18 years, diabetes duration of 27 ± 12 years, and median hemoglobin A1c of 7.1% (6.9, 7.5), using 0.72 (0.4, 1.2) units/(kg·day). For Novolog and Fiasp, respectively, the %TIR (70–180 mg/dL) was 75.3 ± 9.5 and 78.4 ± 9.3; %time <70 mg/dL was 3.1 ± 2.1 and 2.3 ± 2.0; %time >180 mg/dL was 21.6 ± 9.0 and 19.3 ± 8.9; mean glucose was 147 ± 12 and 146 ± 12 mg/dL; coefficient of variation was 28.6% ± 4.5% and 26.8% ± 4.4%; %time in auto mode 86.4 ± 9.2 and 84.4 ± 9.2. All comparisons were nonsignificant for insulin type. Total daily dose (Novolog 48.8 ± 28.4 vs. Fiasp 52.4 ± 31.7 units; P = 0.01) and daily basal (Novolog 17.6 [15.5, 33.8] vs. Fiasp 19.1 [15.3, 38.5] units; P = 0.07) correlated with TIR and %time >180 mg/dL. For insulin delivery in auto mode there was no statistical difference in total daily dose or daily basal between arms. Paired analysis for matched breakfast meals revealed no significant differences in time to maximum glucose, peak glucose, or glucose excursion.

Conclusions:

In this pilot study, the use of either Novolog or Fiasp in a commercially available MiniMed 670G system operating in auto mode resulted in clinically similar glycemic outcomes, with a slight increase in daily insulin requirements using Fiasp.

Introduction

The first commercial hybrid closed-loop (HCL) system became available in 2016 after clinical trials revealed increased time in range (TIR) with a reduction in hypoglycemia for participants with type 1 diabetes.¹ HCL technology offers a promising step forward in reaching clinical goals of diabetes management, and multiple studies have shown improved TIR without an increased risk of hypoglycemia.² Current automated insulin delivery technology is, however, constrained by the pharmacokinetics of current subcutaneously delivered insulin analogs.³ Latency in peak serum insulin levels prevents instantaneous reduction of a rising blood glucose, and the extended duration of insulin action can result in hypoglycemia hours after initial insulin delivery.

Faster-acting insulin aspart, Fiasp^® (NovoNordisk, Bagsværd, Denmark), offers a more rapid onset of insulin action when compared with current rapid-acting insulin analogs. Fiasp is insulin aspart formulated with two well-known excipients generally recognized as safe, niacinamide and l-arginine. Niacinamide is considered responsible for faster initial absorption after subcutaneous administration, and l-arginine serves as a stabilizing agent. A study where Fiasp was administered by subcutaneous injection showed twice-as-fast onset of appearance, a twofold higher early exposure, and >50% greater early glucose-lowering effect compared with Novolog^®.^4,5 When Fiasp is administered by subcutaneous insulin infusion pump therapy, the rapid onset of action is even more pronounced and the time to half-maximal activity is reduced by 11.8 min, the time to peak activity is reduced by 25.7 min, and the duration of insulin activity is reduced by 35.4 min when compared with Novolog.⁶

The MiniMed™ 670G (Medtronic, Northridge, CA) HCL system automatically adjusts basal rates throughout the day and night to minimize hyper- and hypoglycemia. The use of the MiniMed 670G system provides an opportunity to assess how Fiasp works in an HCL automated insulin delivery system. This study was conducted to determine if the use of Fiasp made a clinical difference in real-world performance that was also detectable by the participants or physicians.

Research Design and Methods

This randomized crossover double-blinded pilot study compared the use of Fiasp versus Novolog in an HCL system. The study was approved by the Stanford Human Subjects Research Compliance Office and was conducted in compliance with the standard of Good Clinical Practice and the Declaration of Helsinki. The study is registered on ClinicalTrials.gov (NCT #03554486).

Adult participants (age ≥18 years) were eligible if using any insulin aspart formulation in a MiniMed 670G pump for at least 1 month with a total daily dose of at least 0.3 units/(kg·day), hemoglobin A1c (HbA1c) between 6% and 10%, and consuming at least 60 g of carbohydrates a day. Participants were excluded if they had a hypoglycemic seizure or severe episode of diabetic ketoacidosis in the 6 months before screening, active proliferative diabetic retinopathy, or presence of other significant medical conditions such as adrenal disorders and renal failure. Written informed consent was obtained from all participants before performing any study-specific procedures.

Potential participants were evaluated for study eligibility through elicitation of a medical history, physical examination, and a point-of-care HbA1c. Eligible participants were then trained on study procedures, including proper use of the glucose sensor, blood glucose and ketone monitoring, and how to determine and document any infusion set failures. Participants had a 2-week run-in period on their usual home insulin with weekly downloads of their data onto a research CareLink™ account for review by a physician diabetes specialist to optimize glycemic control.

The Stanford research pharmacy randomly assigned participants within blocks of 4–2 weeks of using either Novolog or Fiasp insulin initially and then crossed over to using the other insulin for an additional 2 weeks. Unlabeled glass vials with bromobutyl rubber stoppers were used, and the vials were labeled with the subject id and “treatment #1” or “treatment #2.” The code to the insulin being used for each treatment assignment was kept by the pharmacy until the end of the study. Treatment #1 did not refer to the insulin being used, but to the order of assignment. The insulin was transferred to the unmarked glass vials the day they were dispensed. The pharmacy delivered the vials to the research staff who then dispensed them to the patient.

The first week of insulin use was to allow adaptation of the MiniMed 670G system to the assigned blinded insulin. At the end of each 2-week period of blinded insulin use, participants and investigators were asked to determine which insulin was used in the previous 2 weeks. Participants were asked to upload their pump at the end of each week for the data to be reviewed by the research staff for safety purposes. In an exploratory study to assess prandial insulin pharmacodynamics of Novolog compared with Fiasp, we requested participants eat the same breakfast for 3 days during the second week of using each insulin. The meals were determined by the participants to be representative of meals they would usually eat. For these replicated meals, they were asked to eat at least 30 g of carbohydrate for breakfast. This constraint was chosen to determine if there was a benefit with the rapid onset of Fiasp with a higher carbohydrate low fat meal such as breakfast. Conducting meal studies at breakfast avoided the impact of previous meals, physical activity, and recent insulin boluses, and with the closed-loop system, fasting glucose was generally well controlled with minimal nocturnal hypoglycemia. Participants were instructed to deliver all meal boluses at the onset of eating. A survey was sent out after each meal challenge day to record the meal contents and nutritional makeup.

We reviewed each recorded meal to determine which meals to include for pharmacodynamic testing. A breakfast meal was included in analysis if the baseline sensor glucose was in the range of 80–160 mg/dL with a rate of change <1.0 mg/(dL·min) in the 20 min before breakfast, with no subsequent bolus within 2 h.

All comparisons between the insulins were made during the second week that they were used in the MiniMed 670G pump. Primary outcomes were percentage TIR (70–180 mg/dL) and hypoglycemia (<70 mg/dL). Secondary outcomes included percentage time with hyperglycemia (>180 mg/dL), mean glucose, glucose coefficient of variation, total daily insulin dose, total daily basal insulin, number of boluses each day, daily carbohydrates, and percentage time in auto mode. The breakfast pharmacodynamic analysis was an exploratory outcome. For the matched breakfast meals meeting criteria, we assessed the glucose peak, time to glucose peak, time to half maximum glucose, and hypoglycemia in the first hour after the meal bolus. Baseline characteristics were compared using paired t-tests for normally distributed measures and Wilcoxon signed rank tests for non-normal distribution. Results with a normal distribution are given as mean ± standard deviation, and results that did not have a normal distribution are reported as median and interquartile ranges (IQR, representing the 25th and 75th percentiles). Baseline characteristics and outcome measures were compared using paired t-tests for normally distributed measures and Wilcoxon signed rank tests for paired samples with non-normal distribution. Data were analyzed using a linear mixed model (SAS statistical software, version 9.4; SAS Institute, Inc., Cary, NC) with treatment (Fiasp or Novolog), period (randomization order), total daily dose of insulin, basal insulin delivery, number of boluses and carbohydrate intake as fixed variables; %TIR, %time <70 mg/dL, %time >180 mg/dL, mean glucose, glucose variability, and time in auto mode as dependent variables; and the subject as a random effect.

Results

A total of 43 individuals were approached to participate in the study through a local diabetes research database or during a clinic appointment. Of the 43, 27 agreed to enroll, with baseline demographics presented in Table 1. Four participants did not meet eligibility criteria and four were lost-to-follow-up. These groups did not differ significantly from those completing the study. Three male participants who did not complete the study were between 19 and 27 years at the time of consent. These individuals were truly lost to follow-up, with multiple failed attempts made at contact. One female volunteer, age 63 years, was consented but then changed her mind about participating in the study.

Table 1.

Baseline Demographic Data

	Completed	Screen failure	Lost to follow-up
n	19	4	4
Age (years)	40.4 ± 17.7	39.6 ± 21.7	33.9 ± 19.7
Diabetes duration (years)	26.6 ± 12.3	20.9 ± 13.7	17.5 ± 11.1
Gender
Male	10 (53%)	2 (50%)	3 (75%)
Female	9 (47%)	2 (50%)	1 (25%)
HbA1c	7.1% (6.9, 7.5)	8.4% (7.4, 9.5)	8.0% (7.2, 8.9)
Total daily dose units/(kg·day)	0.72 (0.4, 1.2)	0.99 (0.78, 1.2)	0.84 (0.6, 1.1)

HbA1c, hemoglobin A1C.

Nineteen participants (10 male and 9 female) were randomized and completed the crossover study. Mean age was 40.4 ± 17.7 years (range 18–76) and mean duration of diabetes was 26.6 ± 12.3 years (range 9.4–53.3). On admission to the study, their median (IQR) insulin dose was 0.72 units/(kg·day) (0.4, 1.2) and HbA1c was 7.1% (6.9, 7.5). Four changes to insulin to carbohydrate ratios and three adjustments to basal insulin profiles were made during the 2-week run-in period.

A modal day for the second week of insulin use is depicted in Figure 1. Differences in dependent variables for the mixed model analysis are summarized in Table 2. Differences in glycemic outcomes were nonsignificant between Novolog and Fiasp, although TIR nearly reached significant with P = 0.0509. The participants correctly determined which insulin they were using 53% of the time and the diabetologists were correct 58% of the time.

FIG. 1.

Median (solid line) and IQR (shaded region) of CGM data for all subjects on the second week of insulin use. Novolog^® is depicted in red and Fiasp^® in blue. Horizontal lines delineate “in range” values of 70–180 mg/dL. CGM, continuous glucose monitoring; IQR, interquartile range.

Table 2.

Differences in Dependent Variables Between Novolog and Fiasp

Dependent variable	Novolog^® ^, ^a	Fiasp^® ^, ^a	Difference (CI)	P
% time in range (70–180 mg/dL)	75.3 ± 9.5	78.4 ± 9.3	−3.1 (−6.3 to 0.02)	0.051
% time <70 mg/dL	3.1 ± 2.1	2.3 ± 2.0	0.8 (−0.4 to 2.1)	0.17
% time >180 mg/dL	21.6 ± 9.0	19.3 ± 8.9	2.2 (−1.1 to 5.6)	0.18
Glucose (mg/dL)	147 ± 12	146 ± 12	0.9 (−4.5 to 6.3)	0.74
% glucose coefficient of variation	28.6 ± 4.5	26.8 ± 4.4	1.8 (−0.1 to 3.7)	0.07
% time in auto mode	86.4 ± 9.2	84.4 ± 9.2	2.0 (−4.3 to 8.3)	0.52

Least squares mean ± SD.

SD, standard deviation.

Insulin delivery data for the second week of insulin use are depicted in Figure 2 and summarized in Table 3. Total daily insulin dose during each week was significantly different (P = 0.01) for Novolog versus Fiasp (48.8 ± 28.4 and 52.4 ± 31.7 units, respectively). Basal delivery during each week, including adjustment made by auto mode, was comparable (P = 0.070) between Novolog (17.6 [15.5, 33.8] units) and Fiasp (19.1 [15.3, 38.5] units). There was no difference in the average number of daily boluses (6.8 ± 2.3 for Novolog and 6.9 ± 2.0 for Fiasp) or in the mean daily carbohydrates as entered in the bolus calculator (152 [117, 198)] for Novolog and 153 [122, 207] for Fiasp) between the two arms of the study during these weeks. Overall, there was similar use of auto mode for each insulin (LS-mean 86.4% ± 9.2% for Novolog and 84.4% ± 9.2% for Fiasp). There were, however, individuals where auto mode use varied significantly between the two arms of the study (for instance, one subject had 94% use of auto mode during the second week of Novolog use and 62% during the second week of Fiasp use). To better characterize the differences in insulin delivery we performed an analysis isolated to auto mode. In this analysis, there was no difference in total daily insulin or daily basal (Table 3). The amount of insulin used for carbohydrate coverage (23.7 ± 14.5 units/day for Novolog and 23.5 ± 14.7 units/day for Fiasp) and for correction doses (3.0 ± 2.1 units/day for Novolog and 2.8 ± 1.9 units/day for Fiasp) was also not different for the two insulins in auto mode.

FIG. 2.

Median (solid line) and IQR (shaded region) of insulin delivery from auto mode for all subjects on the second week of insulin use. Novolog is depicted in red and Fiasp in blue.

Table 3.

Differences in Insulin Delivery Between Novolog and Fiasp

Insulin delivery (units/day)	Novolog^a	Fiasp^a	P
Total daily insulin	48.8 ± 28.4	52.4 ± 31.7	0.01
Total daily basal	17.6 (15.5, 33.8)	19.1 (15.3, 38.5)	0.07
Total daily insulin in auto mode	40.1 (24.9, 55.1)	36.0 (31.2, 75.7)	0.18
Total daily basal in auto mode	15.7 (11.4, 29.2)	17.4 (14.8, 30.4)	0.30

Mean ± SD or median (IQR).

IQR, interquartile range.

Within the linear mixed model, total daily dose exhibited a significant relationship with both %TIR (P = 0.0142) and %time >180 mg/dL (P = 0.0231). Basal delivery specifically had a significant impact on %TIR (P = 0.0082) and %time >180 mg/dL (P = 0.0071). When the linear mixed model was applied to the data in auto mode alone, the associations were nonsignificant. None of the remaining fixed variables significantly impacted the dependent variables.

One hundred sixteen breakfast meals were recorded by the 19 participants. Upon review of submitted meal data, 84 breakfast meals (72% of total) were excluded for the following reasons: 36 (43%) did not have matched meals for both arms (Novolog and Fiasp); 19 (23%) had multiple boluses for breakfast (9 with Novolog and 10 with Fiasp); 9 (11%) had inconsistent carbohydrates entered for meal boluses between the 2 arms; 8 (10%) had missing continuous glucose monitoring (CGM) data; 6 (7%) had missed meal boluses; 5 (6%) had unstable CGM data at baseline; and 1 (1%) was not in auto mode at the time of the bolus. We ultimately included 32 breakfast meals for the pharmacodynamic analysis. The results of this paired meal analysis are given in Table 4 and glucose change from time of bolus are plotted in Figure 3. There were no sensor values <70 mg/dL in the 1 h after meal boluses, and no clinically reported hypoglycemia. The macronutrient content of the breakfast meals used in the analysis were similar for both insulin arms: 41.1 ± 22.2 g carbohydrate, 11.5 ± 11.8 g fat, and 14.3 ± 17.3 g protein for Novolog; 40.9 ± 22.5 g carbohydrate, 11.3 ± 11.6 g fat, and 14.6 ± 16.7 g protein for Fiasp.

FIG. 3.

Median (solid line) and IQR (shaded region) of glucose deviation from time of bolus for selected paired breakfast meals. Novolog is depicted in red and Fiasp in blue.

Table 4.

Differences in Pharmacodynamic Outcomes Between Paired Novolog and Fiasp Breakfast Meals

	Paired breakfast meals (n = 32)
Outcome	Novolog ^a (n = 16)	Fiasp ^a (n = 16)	Paired difference	P
Time to peak (min)	73 ± 23	77 ± 34	4 ± 39	0.68
Time to half maximum (min)	34 ± 15	42 ± 25	8 ± 32	0.32
Peak glucose (mg/dL)	202 ± 40	191 ± 26	−11 ± 46	0.36
Glucose excursion (mg/dL)	73 ± 30	66 ± 34	−6 ± 44	0.58

Mean ± SD.

There were no diabetic ketoacidosis, severe hypoglycemia, or other serious adverse event episodes during the trial; no reported early infusion set failures in the group assigned to Fiasp; and two reported early infusion set failures in the group assigned to Novolog. Of the early infusion set failures, one was due to the infusion set kinking and another was due to local irritation/inflammation at the infusion site. The set failure due to infusion set kinking resulted in an admittance to the emergency room for hyperglycemia with blood sugars >500 and ketones of 2.5 mmol/L but did not meet criteria for diabetic ketoacidosis.

Discussion

In this pilot study, Fiasp use in the MiniMed 670G system did not demonstrate a significant glycemic benefit over Novolog. The 3.1% difference in TIR did not quite reach statistical significance (P = 0.051), and different results may have been seen with a larger clinical study. It is generally felt that a clinically significant change in TIR is 5%^7
–9; however, given our participant's well-controlled baseline TIR of 75%, a 5% improvement is more difficult to achieve.

There was a statistically significant increase in the mean total daily insulin dose used during the second week of Novolog use compared with the second week of Fiasp use. There was no difference in the total daily basal, number of daily boluses, and daily carbohydrate intake between Novolog and Fiasp. To understand the etiology of the differences in total daily insulin usage, we did an additional analysis only evaluating the time participants were in auto mode. When in auto mode, there was no difference in the total daily dose, total daily basal, and bolus insulin between the Novolog and Fiasp arms. We, therefore, conclude that the difference in total daily insulin was attributable to insulin delivery while out of auto mode.

The MiniMed 670G HCL algorithm includes adaptive controller gains and insulin constraints, which allow for some compensation in variability due to insulin sensitivity, meal absorption, and insulin pharmacokinetics. The algorithm, controller gains, and insulin constraints were not specifically modified to account for the pharmacokinetic properties of Fiasp, which might further improve glycemic outcomes among users of Fiasp.

By their nature, HCL systems require users to announce meals. Under ideal conditions, where all factors are replicated as closely as possible, Fiasp demonstrates reduced postprandial glucose excursions.^6,10 Under real-world conditions, individuals may bolus at different times and eat at different rates despite requests to eat the same meal and bolus beforehand. These seemingly minor differences can counteract the benefits of a slightly faster-acting insulin. Differences in insulin kinetics might have a more pronounced effect with fully automated insulin delivery that does not require meal announcement and could reduce variability from user-initiated boluses.

For the purposes of fully automated insulin delivery, an instantaneous on-and-off behavior would be ideal to reduce glucose excursions and prevent hypoglycemia in real time. An earlier onset of insulin action could address prandial glucose rises and decreasing the residual insulin after a meal could decrease late postprandial hypoglycemia. Fully automated insulin delivery algorithms (such as OpenAPS oref1 or DreaMed “GlucoSitter”) deliver insulin without user input. However, a previous double-blinded randomized crossover trial of the DreaMed full closed-loop system did not find significant differences between the use of Novolog and Fiasp in glucose metrics nor matched meals in a controlled environment.¹¹ In this system, the model of insulin pharmacokinetics and pharmacodynamics were also not optimized to fit the actions of Fiasp. If these systems are to derive benefit from Fiasp, it appears they will, at minimum, also need to maintain internal models of insulin pharmacokinetics and/or pharmacodynamics.

Fiasp has demonstrated reduction in HbA1c and postprandial glucose in individuals with type 1 diabetes using multiple daily injections.^4,5,12
–14 The earlier onset of action provides more flexibility in timing of mealtime dosing. Recent studies of Fiasp use in standard insulin pump therapy in response to standardized meal tests support an improved postprandial effect. A double-blind randomized 16-week clinical trial with 472 adults with type 1 diabetes, the Onset 5 trial, found superiority of Fiasp over Novolog in postprandial glucose levels at 30 min, 1, and 2 h postmeals.¹⁵ These findings supported the results of a prior double-blind crossover clinical trials.^10,16 However, neither demonstrated significant benefit in overall median interstitial glucose levels or percentage time in hypoglycemia.^10,15 Based on the Onset 5 Trial,¹⁵ the Food and Drug Administration approved the use of Fiasp in insulin pumps in October 2019.¹⁷

The strengths of this study include the double-blinded randomized crossover design and real-world conditions. The data demonstrated that Fiasp could be safely used with the MiniMed 670G system in an outpatient setting without any significant infusion site issues. The free-range conditions of the trial also served as a limitation. Many subjects were inconsistent in following the meal study requirements, thus decreasing the number of valid matched meals for analysis. Although participants were instructed to deliver their meal boluses in both study arms immediately before the start of the meal, we cannot confirm the timing of all boluses relative to meal start time. This was a pilot study and not statistically powered. Other external factors common in everyday diabetes management, such as treatment of lows and variable eating hours, added additional noise to the data. Additional limitations of the study included the short duration with a small population size. Another three-way random-order crossover trial among 34 adults compared 7 days use of Humalog^® or Novolog with Fiasp in an HCL pump and revealed no significant difference in mean glucose (P = 0.64) or TIR (P = 0.54).¹⁸ Our study and the other two Fiasp closed-loop studies had relatively small sample sizes and short duration. This groundwork sets the stage for a longer trial with a larger sample size to draw further conclusions regarding the use of Fiasp with automated insulin delivery.

In summary, Fiasp did not show a significant advantage over Novolog when used with a commercially available MiniMed 670G system in auto mode. The study findings suggest that Fiasp can be used safely in an HCL system without increased risk for hypoglycemia. Further optimization of the control algorithm and with modeling of the pharmacodynamic and pharmacokinetic properties of Fiasp, or a larger and/or longer clinical trial with more variable baseline glycemic control, may be necessary to determine if there is a clinically significant improvement in glycemic outcomes with use of Fiasp. Additional trials of other forms of ultrarapid acting insulins,¹⁹ including ultrarapid lispro (LY900014) and BioChaperone Lispro,^20,21 are underway, indicating great interest in developing faster-acting insulins as part of standard care. As ultrarapid acting insulins are becoming more prevalent in diabetes management, further research is required to optimize closed-loop algorithms for these novel insulins.

Authors' Contributions

L.H. wrote the article and coordinated the study. B.B. wrote the original protocol, obtained IRB approval, revised the article, and served as principal investigator. M.B. revised the article and helped recruiting patients. L.E. revised the article and enrolled patients. R.v.E. performed the statistical analysis. J.W. performed meal data analysis. R.A.L. revised the article, helped recruit patients, performed data validation, and generated figures. B.B. is the guarantor of the study.

Disclaimer

The data content is solely the responsibility of the authors and does not necessarily represent the official views of NIH.

Footnotes

Author Disclosure Statement

L.H., M.B., L.E., R.v.E., and J.W. have no conflicts of interest or disclosures; B.B. is on medical advisory boards for Convatec, Medtronic, and Tolerion; R.A.L. has consulted for GlySens Incorporated, Abbott Diabetes Care, Biolinq, Capillary Biomedical, and Tidepool.

Funding Information

This study was supported by an investigator-initiated research grant funded by the External Research Program of Medtronic, manufacturer of MiniMed 670G. The REDCap platform services are made possible by Stanford School of Medicine Research Office. The REDCap platform services at Stanford are subsidized by the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health, through grant UL1 TR001085. B.B. has received research support from Medtronic, Tandem, Insulet, Dexcom, NIH (DP3 DK104059, DP3 DK101055, DK-14-024), Helmsley Foundation, and JDRF. L.E. is supported by a Diabetes, Endocrinology and Metabolism Training Grant (1K12DK122550) from NIDDK and has additional research support from the Stanford Maternal and Child Health Research Institute. R.A.L. is supported by a Diabetes, Endocrinology and Metabolism Training Grant (1K12DK122550) from NIDDK and has additional research support from the Stanford Maternal and Child Health Research Institute.

References

Garg

, Weinzimer

, Tamborlane

, et al.: Glucose outcomes with the in-home use of a hybrid closed-loop insulin delivery system in adolescents and adults with type 1 diabetes. Diabetes Technol Ther, 2017; 19:155–163.

Weisman

, Bai

, Cardinez

, et al.: Effect of artificial pancreas systems on glycaemic control in patients with type 1 diabetes: a systematic review and meta-analysis of outpatient randomised controlled trials. Lancet Diabetes Endocrinol, 2017; 5:501–512.

Lal

, Ekhlaspour

, Hood

, Buckingham

: Realizing a closed-loop (Artificial Pancreas) system for the treatment of type 1 diabetes. Endocr Rev, 2019; 40:1521–1546.

Heise

, Hovelmann

, Brondsted

, et al.: Faster-acting insulin aspart: earlier onset of appearance and greater early pharmacokinetic and pharmacodynamic effects than insulin aspart. Diabetes Obes Metab, 2015; 17:682–688.

Hovelmann

, Heise

, Nosek

, et al.: Pharmacokinetic properties of fast-acting insulin aspart administered in different subcutaneous injection regions. Clin Drug Investig, 2017; 37:503–509.

Heise

, Zijlstra

, Nosek

, et al.: Pharmacological properties of faster-acting insulin aspart vs insulin aspart in patients with type 1 diabetes receiving continuous subcutaneous insulin infusion: a randomized, double-blind, crossover trial. Diabetes Obes Metab, 2017; 19:208–215.

Vigersky

, McMahon

: The relationship of hemoglobin A1C to time-in-range in patients with diabetes. Diabetes Technol Ther, 2019; 21:81–85.

Beck

, Bergenstal

, Cheng

, et al.: The relationships between time in range, hyperglycemia metrics, and HbA1c. J Diabetes Sci Technol, 2019; 13:614–626.

Battelino

, Danne

, Bergenstal

, et al.: Clinical targets for continuous glucose monitoring data interpretation: Recommendations from the international consensus on time in range. Diabetes Care, 2019; 42:1593–1603.

10.

Bode

, Johnson

, Hyveled

, et al.: Improved postprandial glycemic control with faster-acting insulin aspart in patients with type 1 diabetes using continuous subcutaneous insulin infusion. Diabetes Technol Ther, 2017; 19:25–33.

11.

Dovc

, Piona

, Yesiltepe Mutlu

, et al.: Faster compared with standard insulin aspart during day-and-night fully closed-loop insulin therapy in type 1 diabetes: A double-blind randomized crossover trial. Diabetes Care, 2020; 43:29–36.

12.

Mathieu

, Bode

, Franek

, et al.: Efficacy and safety of fast-acting insulin aspart in comparison with insulin aspart in type 1 diabetes (onset 1): a 52-week, randomized, treat-to-target, phase III trial. Diabetes Obes Metab, 2018; 20:1148–1155.

13.

Russell-Jones

, Bode

, De Block

, et al.: Fast-acting insulin aspart improves glycemic control in basal-bolus treatment for type 1 diabetes: results of a 26-week multicenter, active-controlled, treat-to-target, randomized, parallel-group trial (onset 1). Diabetes Care, 2017; 40:943–950.

14.

Rose

, Kadowaki

, Pieber

, et al.: Efficacy and safety of fast-acting insulin aspart in people with type 1 diabetes using carbohydrate counting: a post hoc analysis of two randomised controlled trials. Diabetes Ther, 2019; 10:1029–1041.

15.

Klonoff

, Evans

, Lane

, et al. A randomized, multicentre trial evaluating the efficacy and safety of fast-acting insulin aspart in continuous subcutaneous insulin infusion in adults with type 1 diabetes (onset 5). Diabetes Obes Metab, 2019; 21:961–967.

16.

Zijlstra

, Demissie

, Graungaard

, et al.: Investigation of pump compatibility of fast-acting insulin aspart in subjects with type 1 diabetes. J Diabetes Sci Technol, 2018; 12:145–151.

17.

Tucker

: FDA OKs Faster-Acting Insulin Aspart for Pump

Use

. https://www.medscape.com/viewarticle/920162 (accessed March 1, 2020 ).

18.

Russell

: Use of the Ultra-Rapid Insulin Fiasp in the iLet Bionic Pancreas. North Bethesda, MD: Diabetes Technology Meeting, 2019.

19.

Owens

, Bolli

: The continuing quest for better subcutaneously administered prandial insulins: a review of recent developments and potential clinical implications. Diabetes Obes Metab, 2020; 22:743–754.

20.

Andersen

, Meiffren

, Lamers

, et al.: Ultra-rapid BioChaperone Lispro improves postprandial blood glucose excursions vs insulin lispro in a 14-day crossover treatment study in people with type 1 diabetes. Diabetes Obes Metab, 2018; 20:2627–2632.

21.

Heise

, Meiffren

, Alluis

, et al.: BioChaperone Lispro versus faster aspart and insulin aspart in patients with type 1 diabetes using continuous subcutaneous insulin infusion: a randomized euglycemic clamp study. Diabetes Obes Metab, 2019; 21:1066–1070.