Fast‐acting insulin aspart in people with type 2 diabetes: Earlier onset and greater initial exposure and glucose‐lowering effect compared with insulin aspart

Abstract Aims To investigate the pharmacokinetic/pharmacodynamic properties of fast‐acting insulin aspart (faster aspart) versus insulin aspart (IAsp) in people with type 2 diabetes (T2D). Materials and methods In a randomized, double‐blind, crossover design, 61 people with T2D usually treated with insulin ± oral antidiabetic drug(s) received single‐dose faster aspart and IAsp (0.3 U/kg) on separate visits. Blood samples for pharmacokinetic assessment were collected frequently until 12 hours post‐dose. Glucose‐lowering effect was determined in a euglycaemic clamp lasting up to 12 hours post‐dose (target 5.0 mmol/L). Results The serum IAsp pharmacokinetic profile and glucose‐lowering effect profile were shifted to the left for faster aspart versus IAsp. Least squares mean (± SE) onset of appearance was 3.3 ± 0.3 minutes for faster aspart, which was 1.2 minutes earlier than for IAsp (95% confidence interval [CI] −1.8;−0.5; P = .001). Onset of action for faster aspart was 8.9 minutes earlier (95% CI −12.1;−5.7; P < .001) than for IAsp. During the first 30 minutes after dosing, 89% larger IAsp exposure (ratio faster aspart/IAsp 1.89 [95% CI 1.56;2.28]; P < .001) and 147% greater glucose‐lowering effect (2.47 [95% CI 1.58;6.22]; P < .001) were observed for faster aspart compared with IAsp. Offset of exposure (time to 50% of maximum IAsp concentration in the late part of the pharmacokinetic profile) occurred earlier for faster aspart (difference faster aspart – IAsp −36.4 minutes [95% CI −55.3;−17.6]; P < .001). The treatment difference of faster aspart – IAsp in offset of glucose‐lowering effect (time to 50% of maximum glucose infusion rate in the late part of the glucose infusion rate profile) was −14.4 minutes (95% CI −34.4;5.5; P = .152). Conclusions In people with T2D, faster aspart was associated with earlier onset and greater initial exposure and glucose‐lowering effect compared with IAsp, as previously shown in people with type 1 diabetes.


| INTRODUCTION
Fast-acting insulin aspart (faster aspart) is a novel insulin aspart (IAsp) formulation with two additional excipients: L-arginine to ensure a stable formulation and niacinamide to provide increased early absorption after subcutaneous dosing 1 ; thus, faster aspart is an insulin with ultrafast action to enable improved postprandial glycaemic control compared to that obtained with previously developed rapid-acting insulin products. 2,3 In people with type 1 diabetes (T1D), faster aspart has an accelerated onset of appearance and an up to twofold larger initial insulin exposure and glucose-lowering effect compared with IAsp. [4][5][6][7] So far, the pharmacokinetic/pharmacodynamic properties of faster aspart have not been investigated in people with type 2 diabetes (T2D).
As T2D is a progressive disease, most patients will eventually need insulin to achieve normoglycaemia. When oral antidiabetic drugs (OADs) alone or in combination with glucagon-like peptide-1 receptor agonists are no longer sufficient to achieve or maintain glycaemic control, current diabetes guidelines recommend the addition of basal insulin. 8 Treatment can be further intensified by the addition of mealtime insulin in a basal-bolus regimen to address postprandial glucose control. 8 It has been shown that early and intensive intervention to control blood glucose lowers the risk of complications related to diabetes. 9 The use of bolus insulin therapy is therefore a necessity in many people with T2D. Consequently, it is relevant to determine the pharmacological characteristics of a new mealtime insulin not only in people with T1D but also those with T2D. 10 The aim of the present study, therefore, was to investigate the pharmacokinetic/pharmacodynamic characteristics of faster aspart versus IAsp for the first time in people with T2D. Graz reviewed and approved the trial protocol. All participants provided written informed consent before any trial-related activities were initiated. The trial was registered at ClinicalTrials.gov (NCT02933853).

| Procedures
Participants attended a screening visit, an OAD washout visit (14-21 days prior to the first dosing visit), two dosing visits (separated by 7-42 days) and a follow-up visit (7-21 days after the last dosing visit). The OAD washout visit was only performed in participants being treated with insulin secretagogues, DPP-4 inhibitors or SGLT2 inhibitors in combination with insulin with or without metformin. At the OAD washout visit, participants were asked to continue any metformin therapy at an unchanged dose throughout the trial and to terminate treatment with any other OAD(s).
At the two dosing visits, participants received single dosing of 0.3 U/kg faster aspart (100 U/mL; Novo Nordisk, Bagsvaerd, Denmark) or IAsp (NovoRapid ® 100 U/mL; Novo Nordisk) in a randomized sequence. Both trial products were provided in blinded PDS290 peninjector prefilled pens (Novo Nordisk) and administered subcutaneously in a lifted skin fold of the lower abdominal wall above the inguinal area. Use of current insulin was terminated in due time to allow washout before faster aspart and IAsp administration.
Participants arrived at the clinical site at 4:30 PM the day before dosing, were served a standardized meal and started fasting from 7:00 PM. In order to assess the pharmacodynamics of faster aspart and IAsp, a euglycaemic glucose clamp was conducted with an overnight run-in period starting at 10:00 PM. Participants received a variable intravenous infusion of human insulin [40 IU Actrapid ® 100 IU/mL (Novo Nordisk) in 99.6 mL saline] or 20% glucose to obtain the plasma glucose (PG) target concentration of 5.0 mmol/L. The trial product was administered between 8:00 AM and 10:00 AM the next morning after PG had stabilized for ≥1 hour with minimum insulin infusion and no glucose infusion. At the time of dosing, any infusion of intravenous insulin was terminated. After PG had declined by 0.3 mmol/L (defined as onset of action), a variable intravenous glucose infusion was initiated to keep PG at the target throughout the clamp. The clamp continued for up to 12 hours after dosing, but was terminated if PG was consistently ≥11.1 mmol/L, with no requirement for intravenous glucose infusion during the previous 30 minutes. The individual clamps were performed with similar high quality across both treatments ( Figure S1 and Table S1 in Appendix S1). 11 Blood sampling for pharmacokinetics was performed frequently from 2 minutes prior to dosing until 12 hours post-dose (Table S2 in Appendix S1).

| Endpoints
All pharmacokinetic and glucose infusion rate (GIR) endpoints to assess onset of exposure and glucose-lowering effect, initial IAsp exposure and glucose-lowering effect, offset of exposure and glucose-lowering effect, and overall exposure and glucose-lowering effect were defined and derived as previously described. 5 In a blinded review of pharmacokinetic profiles, it was observed that selected profiles had pre-dose values above the LLOQ and stayed above the LLOQ for the entire sampling period. It was therefore decided to baseline-correct these profiles using a baseline value derived as the mean of all pre-dose samples. After baseline correction, all pre-dose values and negative values were set to zero. All pharmacokinetic endpoints for the baseline-corrected profiles were derived in the same way as those for the non-baseline-corrected profiles.
The suppression of serum FFA concentration initially after dosing was determined by calculating the area over the baseline-corrected serum FFA concentration-time curve during the first hour (ΔAOC FFA,0-1 h ) and the first 2 hours (ΔAOC FFA,0-2 h ) as well as the time to reach 50% of the maximum FFA decline in the initial part of the FFA profile (t 50% max FFA decline ). Furthermore, the minimum FFA concentration (FFA min ) was derived.

| Statistical analysis
The sample size calculation was based on a comparison of the primary endpoint (area under the curve for serum IAsp during the first 30 minutes after dosing; AUC IAsp,0-30 min ) between faster aspart and IAsp in particpants with T2D, using information for an earlier faster aspart formulation from a previous trial that included both participants with T1D and T2D. 13 A comparison of the secondary endpoint, area under the GIR curve during the first hour after dosing (AUC GIR,0-1 h ), between faster aspart and IAsp, using information from participants with T1D 4 adjusted to the present population of participants with T2D, was also taken into consideration. A total of 56 completers were required to obtain at least 90% power for the detection of a treatment ratio of 1.25 with a within-participant standard deviation of 0.35 on log-scale for AUC GIR,0-1 h . A total of 56 completers resulted in a statistical power of >99% for AUC IAsp,0-30 min assuming a ratio of 2 between faster aspart and IAsp with a within-participant standard deviation of 0.3 on log-scale. To account for participant withdrawals, 60 participants were planned to be randomized. Withdrawn participants or non-evaluable participants could, however, be replaced by additional participants in order to ensure a sufficient number of completed and evaluable participants.
Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, North Carolina). All endpoints were compared between faster aspart and IAsp using a linear mixed model, with treatment and period as fixed effects and participant as a random effect.
The P value for the two-sided test of no treatment difference using a significance level of 5% was estimated from the model. For FFA endpoints, baseline FFA was included as a covariate. Before analysis, AUC IAsp endpoints, maximum IAsp concentration (C max ), AUC GIR,0-1.5 h , AUC GIR,0-2 h and maximum GIR (GIR max ) were log-transformed. Least squares means for each treatment, treatment ratios and 95% confidence intervals (CIs) were calculated on the original scale.
All other endpoints were analysed on a linear scale: least squares means for each treatment, treatment differences and 95% CIs were estimated. Treatment ratios and 95% CIs were determined according to Fieller's method. 14

| Participant disposition and baseline demographics
A total of 176 individuals were screened and 61 were randomized and exposed to trial product (including 60 participants planned to be randomized plus one replacement participant). In total, 57 participants completed the trial. The safety analysis set comprised all 61 participants exposed to trial product. The full analysis set used for pharmacokinetic/pharmacodynamic analysis included 59 participants.
One participant not fulfilling all inclusion criteria and therefore included in error, and one participant receiving faster aspart at both dosing visits were excluded from the full analysis set. The latter participant was replaced. Participant disposition is shown in Figure S2 in Appendix S1. The 61 exposed participants had a mean ± SD age of 61.8 ± 7.6 years, 26.2% were women and all participants were white. The mean body weight was 87.8 ± 13.9 kg, mean BMI was 29.6 ± 3.2 kg/m 2 , mean duration of diabetes was 20.8 ± 7.8 years, and mean HbA1c was 60 ± 11 mmol/mol (7.6 ± 1.0%). At entry into the trial, 26 participants were only treated with insulin, 19 were treated with insulin + metformin, 13 were treated with insulin + metformin + other OADs (DPP-4 inhibitors and/or SGLT2 inhibitors) and three were treated with insulin + other OADs (DPP-4 inhibitors and/or SGLT2 inhibitors).

| Onset, initial exposure and initial glucoselowering effect
The serum IAsp pharmacokinetic profile and glucose-lowering effect profile were both left-shifted for faster aspart compared with IAsp ( Figure 1), indicating earlier onset and greater initial exposure and glucose-lowering effect versus IAsp. Onset of appearance took place 3.3 minutes after faster aspart administration, which was 1.2 minutes earlier than for IAsp (Table 1). Furthermore, time to 50% of maximum IAsp concentration in the initial part of the pharmacokinetic profile (t Early 50% Cmax ) happened 8.5 minutes earlier and time to maximum IAsp concentration (t max ) happened 16.9 minutes earlier for faster aspart versus IAsp (Table 1). Likewise, onset of action and time to 50% of maximum GIR in the initial part of the GIR profile (t Early 50% GIRmax ) took place 8.9 and 11.8 minutes earlier for faster aspart than for IAsp. Time to maximum GIR (t GIRmax ) did not differ significantly between faster aspart and IAsp (Table 1).
Initial IAsp exposure and initial glucose-lowering effect up to 2 hours were both larger for faster aspart versus IAsp (Figure 2).

| Offset of exposure and glucose-lowering effect
Offset of IAsp exposure happened earlier for faster aspart than for IAsp. This is seen from the shorter time to 50% of maximum IAsp

| Free fatty acids
Faster aspart appeared to suppress FFA more quickly than IAsp, as the mean baseline-corrected serum FFA concentration-time profile  Table 2). The mean minimum FFA concentration was 0.08 ± 0.05 mmol/L for both faster aspart and IAsp.

| Safety
Faster aspart and IAsp were well tolerated, and no safety issues were identified in the present trial. A total of 27 treatment-emergent AEs (16 with faster aspart and 11 with IAsp) were reported. The majority of AEs were mild in intensity and assessed as unlikely to be related to T A B L E 1 Onset of exposure and onset of glucose-lowering effect for fast-acting insulin aspart versus insulin aspart in people with type 2 diabetes The overall left-shift of the pharmacokinetic/pharmacodynamic profiles with faster aspart versus IAsp in the present study in participants with T2D was comparable to that seen in people with T1D. 5,6,15 It has been reported that the absorption kinetics of human insulin are slower in people with T2D than in those with T1D 10,16 ; therefore, it is interesting that absolute values for onset of appearance, t Early 50% Cmax , onset of action and t Early 50% GIRmax for faster aspart in the present study in people with T2D are similar to those seen in people with T1D. 5,6,15  was shown with faster aspart versus IAsp from 3 to 4 hours after a meal in one trial. 21 In contrast, there is also a risk that the offset could occur too fast, which would lead to insufficient levels of circulating insulin during the late postprandial phase. 2 However, this issue may be less relevant in people with T2D because the latter part of the insulin pharmacokinetic profile appears to be generally shifted to the right compared to that in people with T1D. 10,16 A limitation to the accurate understanding of t Early 50% Cmax and t Late 50% Cmax was that C max was significantly higher for faster aspart than for IAsp. 22 Importantly, however, the higher C max for faster aspart would presumably imply an artificial increase in t Early 50% Cmax and t Late 50% Cmax . Thus, if C max had been at the same level for faster aspart and IAsp, t Early 50% Cmax and t Late 50% Cmax might have been even shorter for faster aspart compared to IAsp.
A potential limitation to the generalizability of the present study was the inclusion of people with relatively progressed T2D. Since bolus insulin should be introduced as soon as basal insulin alone is insufficient to maintain glycaemic control, faster aspart is more relevant for patients with shorter duration of diabetes than those included in the present study. 8 Along these lines, it is interesting that a phase III trial has shown that faster aspart, applied with a simple titration algorithm, can be added to a basal-only regimen in people with T2D inadequately controlled on basal insulin plus OAD(s), with satisfactory outcome. 23 In conclusion, the present study shows that in people with T2D, T A B L E 2 Decrease in free fatty acids for fast-acting insulin aspart versus insulin aspart in people with type 2 diabetes

DATA ACCESSIBILITY
Individual participant data will be shared in datasets in a de-identified/anonymized format. Datasets from Novo Nordisk sponsored clinical research completed after 2001 for product indications approved in both the European Union and United States will be available, as well as study protocol and redacted Clinical Study Report, according to Novo Nordisk data-sharing commitments. The data will be available permanently after research completion and approval of product and product use in both the European Union and United States, with no end date. The data may be shared with bona fide researchers submitting a research proposal requesting access to data, for use as approved by the Independent Review Board (IRB) according to the IRB Charter (see novonordisk-trials.com). An access request proposal form and the access criteria can be found at novonordisk-trials.com.
The data will be made available on a specialized SAS data platform.