Elucidating the Mechanism of Absorption of Fast-Acting Insulin Aspart: The Role of Niacinamide

Purpose Fast-acting insulin aspart (faster aspart) is a novel formulation of insulin aspart containing two additional excipients: niacinamide, to increase early absorption, and L-arginine, to optimize stability. The aim of this study was to evaluate the impact of niacinamide on insulin aspart absorption and to investigate the mechanism of action underlying the accelerated absorption. Methods The impact of niacinamide was assessed in pharmacokinetic analyses in pigs and humans, small angle X-ray scattering experiments, trans-endothelial transport assays, vascular tension measurements, and subcutaneous blood flow imaging. Results Niacinamide increased the rate of early insulin aspart absorption in pigs, and pharmacokinetic modelling revealed this effect to be most pronounced up to ~30–40 min after injection in humans. Niacinamide increased the relative monomer fraction of insulin aspart by ~35%, and the apparent permeability of insulin aspart across an endothelial cell barrier by ~27%. Niacinamide also induced a concentration-dependent vasorelaxation of porcine arteries, and increased skin perfusion in pigs. Conclusion Niacinamide mediates the acceleration of initial insulin aspart absorption, and the mechanism of action appears to be multifaceted. Niacinamide increases the initial abundance of insulin aspart monomers and transport of insulin aspart after subcutaneous administration, and also mediates a transient, local vasodilatory effect. Electronic supplementary material The online version of this article (10.1007/s11095-019-2578-7) contains supplementary material, which is available to authorized users.


First method for estimation of absorption ratedeconvolution approach
A two-compartment model was fitted to intravenous (i.v.) data from trial NN1218-3949 (n=20) (registered at ClinicalTrials.gov: NCT02089451) (1) using Monolix 2016R1 software (Lixoft, France). The parameters from the two-compartment model were used for an in-house MATLAB script to perform deconvolution on subcutaneous (s.c.) data for fast-acting insulin aspart (faster aspart).
The two-compartment model, with elimination from the central compartment, was parameterized as follows: Where nc is the amount in the central compartment and np is the amount in the peripheral compartment, cc and cp are the corresponding concentrations, V is the total volume of distribution, and FC is the fraction of volume in the central compartment. k is the elimination rate and Vk is the volume of distribution per kg. a(t) is the absorption rate, which is set to 0 for i.v. data.

Derivation of absorption rate
For estimation of the rate of absorption a(t), the first step is to solve the second differential equation using the general solution to a first order linear differential equation: Which has the solution: Where F(x) is the antiderivative of f(x). The second differential equation is: is ∆ ( ), and inserting limits: The solution is: Dividing by and inserting back into the first differential equation the dependency on cp is removed: Rearranging the terms and expressing amounts as concentrations gives: The differentiation of concentration was approximated by the discretization: ( ) ≈ ( 2 ) − ( 1 ) 2 − 1 And the derivatives were padded by duplicating the last derivative.
For each time step, the absorption rate at the next timepoint at2 can be calculated by knowing the parameters from i.v. (Vc,Vp,Δ,k), the current concentration and the next concentration.

Relative absorption rate
The cumulative sum of the absorption rate at yields the total amount of insulin entering the central compartment, which is approximated with the cumulative sum to the last datapoint: The amount in the s.c. depot at time t (SCt) is hypothesized as being the total amount entering the central compartment subtracted by the amount which has already entered: Instead of using the analytical solution, the exponential decay from the s.c. depot was solved using MATLAB's Ordinary Differential Equation solver, where the rates were linearly interpolated. This allowed for a fast setup when testing multiple microboluses with basal infusion (not in this experimental setup). At each time step, the absorption rate relative to the amount in the s.c. depot is calculated by: The relative absorption rate is interpreted as the average absorption rate for a molecule.
profiles with individuals dosed with insulin aspart and faster aspart (2). The deconvolution method assumes that the concentration was 0 at time 0, and data below the lower limit of quantification were removed.

Second method for estimation of absorption ratepopulation modelling approach
In addition to using the deconvolution approach on individual subject profiles, the insulin aspart absorption rate following s.c. administration of faster aspart and insulin aspart was also determined using a (non-linear mixed effects) population modelling approach using NONMEM software (version 7.3; ICON Development Solutions, Ellicott City, MD, USA). The PK data used for this purpose was the i.v and s.c. cross-over data for faster aspart from trial NN1218-3949 (n=20) (1) and the s.c faster aspart and insulin aspart cross-over data from trial NN1218-3978 (n=51) (2). The same data as the first approach. Common to both methods used to estimate the relative absorption rate is that the estimation of remaining insulin aspart in the depot becomes more and more uncertain when the depot becomes small. Since the relative rate of absorption is calculated by using the depot size in the denominator, it follows that there is a limitation in how long time after injection the calculations are reasonble. Hence, we have truncated calculations at 2 hours, at which point only approximately 25% of the injected insulin aspart remains in the depot. Scattering curves corresponding to sample 1 to 4 from top to bottom (see Table II in main manuscript for sample composition). arb., arbitrary; HBSS, Hank's Balanced Salt Solution; SAXS, small angle X-ray scattering.

Niacinamide has a counteracting effect on the Zn 2+ -free oligomerization of insulin aspart
After the dissociation of hexamers, monovalent (Na + ) and divalent ions (Ca +2 , Mg +2 ) present in the s.c. tissue can promote the oligomerization of insulin dimers, further impeding absorption (3,4). The impact of niacinamide on Zn 2+ -independent oligomerization of insulin aspart was investigated using phosphate  arb., arbitrary; ISF, interstitial fluid-like; PBS, phosphate buffered saline; SAXS, small angle X-ray scattering.
The effect of niacinamide and 1-methyl-niacinamide on the Zn 2+ -free oligomerization of insulin aspart in ISF buffer was also investigated with two concentration series. The concentration of insulin aspart was kept constant at 0.6 mM, while the concentrations of the two excipients were varied from 0 to 320 mM. The effect of the two excipients was noticeably different. As can be seen from the recorded SAXS curves in Figure S3, niacinamide had a stronger suppressing effect on oligomerization of insulin aspart compared with 1-methylniacinamide. Moreover, as judged by the maximal value of the P(r)-functions (maximum intramolecular distance, Dmax), 1-methyl-niacinamide appears largely incapable of suppressing the largest oligomers. In contrast, the Dmax values obtained for niacinamide gradually decreased as the concentration of niacinamide increased. Lastly, it should be noted that the result obtained for 1-methyl-niacinamide appears less systematic, which is seen in the plot of the radius of gyration versus concentration ( Figure S3, bottom left panel). It is possible that 1-methyl-niacinamide might have a detrimental impact on the stability of insulin apart, and that this could be the underlying course for the noisier trend observed for this compound.

Niacinamide increases subcutaneous blood flow in pigs
Video: Representative LASCA perfusion experiment.