Subjects were enrolled at Hakata Clinic and included Japanese males aged 20–55 years (both inclusive) with a body mass index of 18.0–28.0 kg/m2, certified as healthy by a comprehensive clinical assessment that included his past medical history, normal vital signs, normal 12-lead electrocardiogram (ECG) parameters and laboratory parameters within the normal ranges.
This was a randomized, single-center, double-blind, single-dose, 2-treatment, 2-period, crossover study undertaken in 2018–2019 (trial registration: www.clinicaltrials.jp, identifier: JapicCTI-205380, registered 20/07/2020). Subjects received each treatment once. The trial was approved by an ethical review board (Hakata Clinic, Fukuoka, Japan) and conducted in accordance with the principles of the Declaration of Helsinki and the International Conference on Harmonization guidelines for Good Clinical Practice. Written informed consent was provided by all subjects before study entry.
Following a screening visit 3–28 days before the first period, subjects enrolled into the study were admitted to the study clinic on the day before dosing. On the day of treatment, they were randomized (computer-generated by sponsor) to one of the two treatment sequences (see Supplementary Fig. S1). Subjects received a single subcutaneous dose of 0.3 U/kg SAR-Asp or NN-Asp-Jp in the first treatment period in randomized order followed by the other drug in the second treatment period. A wash-out period of 7–18 days separated each treatment period, with an end-of-study visit performed 4–8 days after the last dose.
SAR-Asp (100 U/mL) solution for injection was manufactured by Sanofi (Frankfurt, Germany), with NN-Asp-Jp provided as the commercially available formulation. To maintain double-blinding (with respect of the Investigator, subject and Sponsor) and consistency of dosing methodology, an independent pharmacist at the study site was responsible, with an assistant, for preparing the treatments for each subject.
Venous blood samples for pharmacokinetic analysis were collected before dosing of study drug, and then every 15 to 60 min during the 10 h clamp period. Samples were centrifuged within 20 min of collection. Plasma was collected and kept frozen (− 60 to − 80 °C) until analysis. Plasma concentrations of SAR-Asp and NN-Asp-Jp were analyzed by using a validated liquid chromatography-tandem mass spectrometry (LC–MS/MS) assay at Syneos Health (Quebec, Canada). Plasma concentrations within the validated concentration range (100–8000 pg/mL) were used to calculated pharmacokinetic parameters. Inter-assay precision (% coefficient of variation [CV]) and inter-assay accuracy (%bias) during validation were ≤ 15%.
Pharmacodynamic evaluation using euglycemic clamp
Each dosing visit included a 10 h automated euglycemic clamp procedure (STG-55, Artificial Endocrine Pancreas; Nikkiso Co, Ltd, Tokyo, Japan)13. The premise of performing the clamp procedure is that the blood glucose lowering effect of an administered insulin is antagonized by a variable infusion of glucose so that blood glucose is maintained (or clamped) at a target level2,14. The metabolic profile of the investigated insulin is characterized by the glucose infusion rate (GIR) needed to keep blood glucose as close as possible at its predefined target level during the glucose clamp15.
Following an overnight fast of at least 10 h, subjects were connected to the clamp device and their baseline blood glucose was determined (via glucose oxidase sensors that measure whole-blood glucose levels) before dosing of the study drug. For each subject, this was calculated as the mean of four glucose measurements at 30, 20, 10 and 1 min before administration of the study drugs. The clamp procedure was not performed in those subjects having a baseline glucose level less than 70 mg/dL (3.92 mmol/L). After dosing, onset of insulin action in each subject was when the blood glucose dropped below the target level of the clamp procedure, which was when it was 5 mg/dL (0.28 mmol/L) less than their fasting baseline value2.
Blood glucose levels were measured in 1-min intervals, and using a predefined algorithm the clamp device automatically administered a variable infusion of 10% glucose in response to changes in glucose to maintain each subject at their target glucose level. The GIR, being the amount of external glucose needed to keep a subject’s blood glucose concentration at its target level, was continuously measured, and recorded by the STG-55 device. GIR profiles reflected the metabolic effects of SAR-Asp and NN-Asp-Jp over time.
The safety and tolerability of single doses of SAR-Asp and NN-Asp-Jp were assessed by 12-lead ECG, vital signs, routine laboratory parameters, physical examination, and reporting of adverse events (AEs). AEs were classified using MedDRA (Medical Dictionary for Regulatory Activities) 21.1. The safety population included all randomized patients who received at least one dose of study insulin, analyzed according to the treatment received. The treatment-emergent AE period included the time from the first dose of study drug up to 72 h after the last dose in each treatment period.
Pharmacokinetic and pharmacodynamic parameters
The pharmacokinetic analysis dataset included subjects who completed at least one treatment period, had measurable insulin aspart concentrations and no major or critical deviations. Parameter estimates for SAR-Asp and NN-Asp-Jp were calculated by using standard noncompartmental methods with Phoenix WinNonlin 8.1 (Certara, Princeton, NJ). Area under the plasma insulin aspart concentration–time curve was calculated by the trapezoidal method from 0 to the time of the last concentration above the limit of quantification (INS-AUClast) and extrapolated to infinity (INS-AUCinf). Maximum plasma insulin aspart concentration observed (INS-Cmax) and INS-AUClast were the primary pharmacokinetic endpoints of the study. INS-AUCinf was a secondary endpoint.
The GIR over time was the primary pharmacodynamic parameter measured during the clamp procedure. The area under the body weight-standardized GIR time curve [GIR-AUC] measured insulin mediated glucose uptake into tissues. Subjects completing at least one clamp procedure with no major or critical deviations were included in the pharmacodynamic analysis dataset. Individual GIR values in each treatment group were standardized for body weight and subjected to a smoothing procedure using a locally weighted scatterplot smoothing (LOESS) function (SAS, PROC LOESS, factor 0.06). A smoothing factor of 6% was used based on the expected morphology of the GIR-profiles. Individual blood glucose levels were also subjected to a smoothing procedure. Fitted data were used to calculate the pharmacodynamic parameters, including the primary parameters of maximum smoothed body weight standardized GIR (GIRmax) and GIR-AUC over 10 h (GIR-AUC0–10 h), and the secondary parameter of time to reach GIRmax (GIR-tmax).
Parameters used to assess clamp quality included the mean and individual %CV of blood glucose measurements during euglycemia, and the absolute difference of individual mean blood glucose measurements from the clamp target level, as described previously9.
The study aimed to show similarity (equivalence) in pharmacokinetic exposure and glucodynamic activity of a single dose (0.3 U/kg) of SAR-Asp to NN-Asp-Jp under fasting conditions. To achieve this, 26 and 32 evaluable subjects, respectively, were required. These calculations were based on estimates of within-subject variability from a prior SAR-Asp study in subjects with T1D9, and those reported in applicable studies performed in healthy subjects16,17. A within-subject standard deviation (SD) of 0.175 and 0.180 was assumed for the natural log-transformed pharmacokinetic (INS-Cmax and INS-AUClast) and pharmacodynamic (GIRmax and GIR-AUC0–10 h) parameters, respectively, for a true treatment ratio between the two formulations of 0.93 and 1.07, respectively. The planned sample sizes provided at least 90% power to show equivalence with a type 1 error of 5% and 2.5% for the pharmacokinetic and pharmacodynamic parameters, respectively. To allow for dropouts, the study planned to recruit at least 36 individuals (18 per sequence).
Log-transformed pharmacokinetic and pharmacodynamic parameter estimates for SAR-Asp and NN-Asp-Jp were analyzed using a linear mixed-effects model with period, sequence, and treatment as fixed effects and subject as a random effect. For each parameter, the model-based difference in treatment means along with the confidence limits (90% for pharmacokinetic parameters, 95% for pharmacodynamic parameters) was back-transformed to provide estimates for the ratio of geometric means (gMean) between treatments (SAR-Asp/NN-Asp-Jp) and the corresponding confidence limits. Similarity for the pharmacokinetic parameters (INS-AUCinf and INS-Cmax) and bioequipotency for the pharmacodynamic parameters (GIR-AUC0–10 h and GIRmax) was concluded if the confidence limits (90% and 95% confidence intervals [CIs], respectively) of the treatment ratios were entirely within the 0.80–1.25 equivalence interval, in agreement with regulatory guidance2,18. Data was analyzed using SAS v9.4 (SAS Institute, Cary, NC).