A Comparison of Pharmacokinetics and Pharmacodynamics of Insulin Aspart,Biphasic Insulin Aspart 70,Biphasic Insulin Aspart 50,and Human Insulin: A Randomized,Quadruple Crossover Study

Abstract

Background:

We compared the pharmacokinetic and pharmacodynamic profiles of insulin aspart, biphasic insulin aspart 70 (BIAsp70) and 50 (BIAsp50) (containing 70% and 50% rapid-acting insulin aspart, respectively), and soluble human insulin under experimental conditions.

Subjects and Methods:

In this randomized, four-period crossover study, 19 type 1 diabetes patients received subcutaneous injections of identical doses (0.2 U/kg) of insulin aspart, BIAsp70, or BIAsp50 immediately before a standardized meal or human insulin 30 min before meal. Plasma glucose and serum insulin were measured for 12 h postprandially.

Results:

The pharmacokinetic and pharmacodynamic profiles of human insulin differed from those of insulin aspart, BIAsp70, and BIAsp50. The three different aspart preparations had easily distinguishable features with regard to onset and duration of action. Insulin aspart preparations were, on average, absorbed twice as fast as human insulin. In the initial phases (0–4 h and 0–6 h), the insulin area under the concentration–time curve (AUC_ins) was significantly higher during insulin aspart treatment compared with the others, whereas insulin aspart had a significantly lower AUC_ins over the last 6 h (P<0.05). BIAsp70 and BIAsp50 provided insulin coverage comparable to that of human insulin over the last 6 h. Insulin aspart had the most pronounced onset of action and the shortest duration. Comparing with insulin aspart and BIAsp70, BIAsp50 revealed a closer treatment ratio to human insulin on pharmacodynamic end points.

Conclusions:

BIAsp70 and BIAsp50 injected immediately before a meal are at least as effective as human insulin injected 30 min earlier in controlling postprandial glycemic excursions. BIAsp50 showed the greatest similarity to human insulin with regard to pharmacokinetic and pharmacodynamic profiles.

Introduction

E xogenous insulin therapy is necessary for all patients with type 1 diabetes and constitutes an important option for patients with type 2 diabetes to achieve optimal glycemic control.¹ Human insulin regimens have been well established for many years. However, because of the slow absorption, the profile after subcutaneous administration of soluble human insulin fails to mimic the physiological prandial insulin secretions seen in subjects without diabetes,^2,3 even when human insulin is injected the recommended 30 min before a meal.⁴

Insulin aspart (NovoRapid^®, Novo Nordisk, Bagsværd, Denmark), a rapid-acting insulin analog, has been developed to partly overcome the limitations of soluble human insulin with regard to achieving a more physiological prandial insulin profile. By replacing proline with aspartic acid in position B28, insulin aspart results in faster absorption and shorter duration after subcutaneous injection compared with soluble human insulin.⁵ Several studies have demonstrated that insulin aspart provides better control of postprandial glucose excursions than human insulin in both type 1 and type 2 diabetes patients.^6

–9 However, these advantages are barely reflected in the overall glucose control because of the short prandial action of insulin aspart.^7,10,11 In the basal–bolus regimen, which involved human insulin as mealtime insulin and intermediate- or long-acting insulin as basal insulin, human insulin might provide the insulin coverage between meals and basal insulin might only need to be injected at night.¹² Because insulin aspart had a shorter duration of action, an extra intermediate- or long-acting insulin injection during the day time might be necessary in order to cover the basal insulin requirement.^13,14

Biphasic insulin aspart formulations contain different proportions of soluble and protamine-crystallized insulin aspart.^15,16 The protaminated aspart has an intermediate duration of action and provides basal insulin coverage.^15,17 Biphasic insulin aspart 50 (BIAsp50; NovoMix 50^®) and biphasic insulin aspart 70 (BIAsp70, NovoMix 70^®), containing 50% and 70% rapid-acting insulin aspart, respectively, reduces both postprandial and fasting glucose levels by a single injection.¹⁸

There is no doubt that the insulin-producing companies are gradually withdrawing human insulin preparations from the pharmaceutical market¹⁹ and that within a few years only insulin analogs will be available in many areas of the world. However, there are obvious differences in the pharmacokinetic and pharmacodynamic profiles between these insulin preparations, for instance, insulin aspart and human insulin, and for all physicians, it will be a clinical challenge to ensure that human insulin is substituted in the most beneficial way for the patients, in terms of efficacy, safety, and flexibility. Therefore, the aim of this study was to compare the pharmacokinetic and pharmacodynamic profiles of insulin aspart, BIAsp70, BIAsp50, and soluble human insulin for a period of 12 h after a standard meal in patients with type 1 diabetes. In addition, the safety of the four insulin preparations was assessed.

Subjects and Methods

Study design

This was a single-center, open-label, randomized, four-period crossover study. The study was approved by the local Ethics Committee and the Danish Medical Agency and conducted in accordance with the Declaration of Helsinki and the principles of Good Clinical Practice. Written informed consent from all the patients was obtained before any study-related activities.

Subjects

Type 1 diabetes patients were screened and enrolled according to inclusion and exclusion criteria. Inclusion criteria were as follows: age ≥18 years, clinical diagnosis of type 1 diabetes before the age of 40 years, insulin treatment with any regimen ≥12 months, total daily insulin dose ≥0.4 U/kg, glycated hemoglobin of 7–12%, and body mass index of 18–35 kg/m². Exclusion criteria included allergy to insulin, recurrent major hypoglycemia, acute myocardial infarct <2 months or severe heart insufficiency, impaired renal or hepatic function, pregnancy, or breastfeeding.

Procedures

The patients participated in four separate study visits (1 week apart) where they in random order received a subcutaneous injection of identical doses of insulin aspart, BIAsp70, BIAsp50, or human insulin before a standardized meal.

Patients attended the clinical research unit at 9 p.m. on the night prior to the dosing day. They were instructed to inject the last dose of intermediate- or long-acting insulin at least 24 h before administration of the study insulin. Fasting commenced at 10 p.m. and ended at 8 a.m. Intravenous infusions of isotonic glucose and insulin were administered with two individual infusion pumps (IVAC^® model 591 pump; Alaris Medical, Zurich, Switzerland) during the overnight fasting. Soluble human insulin (Actrapid^®, Novo Nordisk) was infused during the night prior to injection of insulin aspart, BIAsp70, or BIAsp50. Conversely, insulin aspart was infused before the injection of human insulin. The intention was to maintain blood glucose values between 5 and 8 mmol/L prior to study insulin injection, with the optimal being 6 mmol/L. A commercially available plasma glucose meter (Ascensia^® Contour; Bayer, Leverkeusen, Germany) was used for blood glucose measurement every 30 min overnight. Before the insulin injection, the glucose values had to be in the defined range for at least 30 min.

In the morning of the study day, a single dose (0.2 U/kg) of either insulin aspart, BIAsp70, BIAsp50, or human insulin (Actrapid) was injected before a standardized meal. Insulin was injected subcutaneously in the abdominal wall. Insulin aspart, BIAsp70, and BIAsp50 were injected immediately before the meal, which started at 8 a.m., whereas human insulin was injected 30 min before the meal. The standardized meal contained 3,360 kJ (15.2% protein, 48.0% carbohydrates, and 35.8% fat). The patients were requested to finish their meal within 15 min. During the study day, levels of plasma glucose and serum insulin were measured every 15 min from 8 a.m. to 4 p.m. and every 20 min from 4 p.m. to 8 p.m. Safety measurements of blood glucose were drawn every hour with the Ascensia Contour plasma glucose meter and more often if needed. The study day ended at 8 p.m. or when the patient's glucose measurement exceeded 16.0 mmol/L, whichever came first.

Hypoglycemic episodes were classified as major (patient is unable to treat episode), minor (blood glucose measurement <3.1 mmol/L), or symptoms only (blood glucose measurement ≥3.1 mmol/L). Episodes were treated with standardized oral glucose (equal to 40 g of glucose) or intravenous injection of 10% glucose.

Laboratory assessment

Serum insulin aspart was measured by specific immunoassays using the luminescent oxygen channeling immunoassay (LOCI™) technology (Assay Technology, Diabetes Research Unit, Novo Nordisk). The lower limit of quantification was 2 pmol/L. Cross-reaction with human insulin was less than 1%. Serum human insulin was measured by a direct sandwich two-side fluoroimmunometric assay with the 1235 AutoDELFIA^® automatic immunoassay system (PerkinElmer Life and Analytical Sciences, Turku, Finland). The lower limit of quantification was 3 pmol/L. Cross-reaction with insulin aspart was less than 2%. Plasma glucose was measured by the immobilized enzyme biosensor on the YSI 2300 STAT Plus™ glucose and lactate analyzer (YSI Inc., Yellow Springs, OH).

Statistical analysis

The study had an 80% power to detect an absolute difference of 2.0 mmol/L between the maximal values of plasma glucose concentrations (Cmax_glu), with an SD of 2.5 mmol/L, suggesting that 16 patients were needed to complete the study. Assuming a dropout rate of approximately 33%, 24 patients were required.

Serum insulin concentrations below the lower limit of quantification were set to 0 pmol/L in the statistical analyses. If a patient discontinued earlier than planned because of hyperglycemia, the serum insulin concentration was set to 0 pmol/L until scheduled termination for the pharmacokinetic analyses. The pharmacodynamic calculations were performed without these results.

Area under the concentration–time curve (AUC) was calculated by the trapezoidal rule and stratified into initial phases (0–4 and 0–6 h) and a late phase (6–12 h). After logarithmic transformation (ln), the differences in Cmax_glu and AUCs were analyzed using analysis of variance with treatment as a fixed factor and patient as a random factor. If the analysis of variance test was significant, the difference between treatments was calculated by using the Bonferroni method. Time was summarized using descriptive statistics. A P value of < 0.05 was considered statistically significant. Statistical analyses were performed using IBM SPSS Statistics version 19.0 (IBM Corp., Somers, NY).

Results

Subjects

In total, 23 patients were randomized, and 19 patients (15 men and four women; all whites) underwent all the study visits. Three patients withdrew for personal reasons, and one withdrew because of lack of compliance. The demographic and baseline characteristics of the 19 patients who completed the study are summarized in Table 1.

Table 1.

Baseline Characteristics of Study Subjects

	Mean (range)
Number of subjects	19
Age (years)	40.7 (22–63)
Duration of diabetes (years)	20.9 (3–47)
Baseline glycated hemoglobin (%)	8.2 (7.0–9.8)
Baseline insulin dose (U/kg)	0.79 (0.49–1.28)
Body mass index (kg/m²)	25.5 (20.6–30.3)

Pharmacokinetics

Figure 1A illustrates the insulin profiles during the 12-h study periods. In the first 4 h, insulin profiles showed clear differences between human insulin and insulin aspart preparations. Compared with human insulin, insulin aspart preparations reached the apparently higher peak concentrations.

FIG. 1.

(A) Serum insulin concentration and (B) plasma glucose concentration during treatment with insulin aspart (open circles), biphasic insulin aspart 70 (black rectangles), biphasic insulin aspart 50 (open rectangles), and human insulin (black circles). The arrow indicates the human insulin injection time, and the vertical dotted line indicates the time of the meal. Serum insulin profiles are given as mean±SEM values, whereas plasma glucose profiles are given as mean values because so few SEM values exceeded 1 mmol/L.

Table 2 demonstrates that insulin aspart preparations were, on average, gave peak values twice as fast as human insulin as assessed by time to maximal concentration. In the initial phases after injection (0–4 and 0–6 h), the insulin AUC (AUC_ins) was significantly higher during insulin aspart treatment compared with the other three insulin treatments (P<0.05). By contrast, insulin aspart had a significantly lower AUC_ins over the last 6 h (P<0.05). No statistical differences were found in AUC_ins when human insulin was compared with BIAsp50 in any interval.

Table 2.

Pharmacokinetic and Pharmacodynamic Results

	Mean (range)				Treatment ratio (95% CI) with human insulin
	Human insulin	BIAsp50	BIAsp70	Insulin aspart	BIAsp50	BIAsp70	Insulin aspart
Pharmacokinetics
Tmax (min)	145 (130–160)^*	69 (54–84)	66 (51–81)	76 (61–91)	—	—	—
AUC_0–4h (pmol·h/L)	645 (595–699)^ab	727 (672–785)^ab	974 (901–1,052)^acd	1,237 (1,146–1,338)^bcd	1.13 (1.04–1.32)	1.51 (1.29–1.76)^d	1.92 (1.64–2.24)^d
AUC_0–6h (pmol·h/L)	893 (829–962)^ab	891 (829–958)^ab	1,156 (1,076–1,242)^acd	1,402 (1,305–1,508)^bcd	1.00 (1.15–1.16)	1.29 (1.12–1.49)^d	1.57 (1.36–1.81)^d
AUC_6–12h (pmol·h/L)	184 (146–232)^a	255 (199–326)^a	195 (152–250)^a	99 (77–127)^bcd	1.39 (0.88–2.17)	1.06 (0.68–1.66)	0.54 (0.34–0.85)^d
	(n=11)	(n=16)	(n=11)	(n=5)
Pharmacodynamics
Cmax (mmol/L)	11.2 (10.2–12.3)^a	11.5 (10.5–12.6)^a	9.8 (8.9–10.8)	8.9 (8.1–9.8)^cd	1.03 (0.82–1.31)	0.89 (0.70–1.12)	0.80 (0.63–1.01)^d
AUC_0–4h (mmol·h/L)	32.7 (29.6–36.0)^a	35.8 (32.4–39.4)^ab	28.8 (26.0–31.8)^c	24.7 (22.3–27.2)^cd	1.09 (0.90–1.32)	0.88 (0.73–1.07)	0.76 (0.62–0.91)^d
AUC_0–6h (mmol·h/L)	45.2 (40.7–50.2)	53.2 (47.9–59.1)^a	43.5 (39.1–48.3)	38.3 (34.6–42.6)^c	1.18 (0.96–1.44)	0.96 (0.78–1.18)	0.85 (0.69–1.04)
AUC_6–12h (mmol·h/L)	52.8 (44.3–62.8)	55.3 (48.5–63.2)	57.4 (48.3–68.4)	64.4 (49.6–83.8)	1.03 (0.60–1.75)	1.14 (0.67–1.94)	1.23 (0.72–2.10)
	(n=11)	(n=16)	(n=11)	(n=5)

The results for maximal concentration of glucose (C _max) (in mmol/L), area under the concentration–time curve (AUC), and time to maximal concentration of insulin (T _max) are expressed as geometric mean (range) drawn from analysis of variance. The ratio refers to analysis of variance comparing insulin aspart, biphasic insulin aspart 70 (BIAsp70), and biphasic insulin aspart 50 (BIAsp50) with human insulin. The analyses for pharmacodynamics are adjusted for baseline plasma concentration of glucose levels at fasting. Each insulin treatment involved 19 patients, unless otherwise noted in parentheses (n indicates the number of patients).

P<0.05 versus insulin aspart; ^b P<0.05 versus BIAsp70; ^c P<0.05 versus BIAsp50; ^d P<0.05 versus human insulin.

Pharmacokinetics data for human insulin obtained from 30 min before a standard meal.

CI, confidence interval.

Pharmacodynamics

Plasma glucose profiles are shown in Figure 1B. In the baseline, fasting plasma glucose levels were comparable among the treatment groups of insulin aspart, BIAsp70, BIAsp50, and human insulin (6.57, 6.99, 6.50, and 6.41 mmol/L, respectively). The different pharmacodynamic effects of the four different insulin preparations are easily distinguishable during the entire study periods.

Table 2 summarizes the postprandial glycemic end points after a standard meal. Results indicated a closely corresponding relation between the pharmacokinetic and pharmacodynamic properties of the four insulin preparations. Insulin aspart preparations with greater proportion of fast-acting component induced lower glucose AUC (AUC_glu) in the initial phase and higher AUC_glu in the later phase. Both insulin aspart and BIAsp70 resulted in a significantly lower AUC_glu compared with BIAsp50 during the first 4 h (P<0.05). In contrast, BIAsp50 had the lower AUC_glu than insulin aspart and BIAsp70 in the later phase. However, no statistical differences were found in this interval. During the first 4 h, human insulin revealed a significantly higher Cmax_glu and AUC_glu compared with insulin aspart (P<0.05). With regard to Cmax_glu and AUC_glu, no statistical differences were found when human insulin was compared with BIAsp70 or BIAsp50 during the 12-h study period. All treatment ratios between BIAsp50 and human insulin were close to 1.0 on pharmacodynamic end points.

Hypoglycemic episodes

Twelve of the 19 patients experienced 36 minor hypoglycemic episodes during the study visits. No major hypoglycemic episodes occurred. The number of hypoglycemic episodes was higher in patients treated with insulin aspart (16 episodes) than patients treated with human insulin (10 episodes), BIAsp70 (seven episodes), and BIAsp50 (three episodes). The mean time of the first intervention after the start of meal was 160 min with insulin aspart, 177 min with BIAsp70, 126 min with BIAsp50, and 236 min with human insulin.

Hyperglycemic episodes

Because of the intolerable hyperglycemia, 14 patients terminated earlier than planned in at least one study day. All the early terminations occurred during the last 6 h. Fourteen cases occurred at a mean (range) after 544 (360–660) min with insulin aspart treatment, eight cases occurred after 610 (560–660) min with BIAsp70, three cases occurred after 593 (560–620) min with BIAsp50, and eight cases occurred after 595 (520–680) min with human insulin.

Discussion

This is the first head-to-head comparison of the pharmacokinetics and pharmacodynamics of soluble human insulin with the insulin analog formulations, insulin aspart, BIAsp70, and BIAsp50 in type 1 diabetes patients. Our results demonstrate that the pharmacokinetic profiles differed among human insulin, insulin aspart, BIAsp70, and BIAsp50; each preparation had distinguished features with regard to onset and duration of action. The differences of the pharmacokinetic profiles are clearly reflected by the pharmacodynamics. The action of insulin aspart appeared early, but the duration was short. Human insulin had the slowest onset and longest duration of action compared with the other preparations in spite of being injected 30 min earlier than the aspart preparations (i.e., 30 min before the meal).

Compared with an equal dose of human insulin, insulin aspart was absorbed twice as fast and was superior on early postprandial glycemic control by having significantly lower Cmax_glu and AUC_glu in the first 4 h. The early onset of insulin aspart provided better control of early postprandial hyperglycemia; however, its short duration also caused insufficient insulin coverage between meals, resulting in elevated preprandial glucose levels. Indeed, we observed that 14 of 19 patients treated with insulin aspart had to discontinue after 6 h because of intolerable hyperglycemia. Although no difference was found in pharmacodynamic end points between insulin aspart and human insulin in the late phase, it could be due to lack of patient data and thus power for detecting changes, as many patients were discontinued because of hyperglycemia.

In the current study, the observed pharmacokinetic and pharmacodynamic profiles of the applied insulin aspart preparations are consistent with previous published results.^15,16 During the initial phase after injection, insulin aspart had significantly higher AUC_ins values, followed by BIAsp70 and BIAsp50. By contrast, AUC_ins of BIAsp50 was highest compared with the others in the late phase. This demonstrates that the pharmacokinetic profiles of soluble aspart and protamine-crystallized insulin aspart are well preserved in various formulations. Because of protamine-crystallized insulin aspart in BIAsp70 and BIAsp50, both preparations provided insulin coverage comparable to that of human insulin over the last 6 h. As we expected, the pharmacokinetic difference was translated into pharmacodynamic profiles. With lowest proportions of soluble aspart, BIAsp50 resulted in the significantly highest AUC_glu in the first 4 h and the lowest glucose levels during the last 6 h.

Direct comparisons between serum concentration of human insulin and insulin aspart (pure insulin aspart, BIAsp70, and BIAsp50) are not rigorous because these insulin preparations are made up of different molecules and are analyzed by different assays. However, the pharmacodynamic measures of the action of insulin are more indicative and clinically relevant than the pharmacokinetic measures. Although BIAsp70 and BIAsp50 differed from human insulin on pharmacokinetic profiles, both had similar capabilities for glucose control in terms of AUC_glu, not only in the initial phase but also in the late phase after injection. With regard to Cmax_glu and AUC_glu, treatment ratios between BIAsp50 and human insulin were close to 1.0 during the study period.

In the current study, 12 patients experienced 36 minor hypoglycemic episodes. The intervention with extra glucose load was supposed to increase the glucose level and the AUC_glu for a certain period, which affected the pharmacodynamic assessment and might also induce hyperglycemia later on. Therefore, efficacy of insulin could be underevaluated because of these interventions. Among those hypoglycemic episodes, 16 occurred during insulin aspart treatment versus 10 episodes with human insulin. In a long-term safety trial comparing premeal insulin aspart with premeal soluble human insulin in 753 patients with type 1 diabetes, Home et al.²⁰ also reported the relative risk of having a minor hypoglycemic episode during the treatment period was significantly higher in the insulin aspart group than in the human insulin group. Nevertheless, a systematic review showed that insulin aspart had a significant reduction in nocturnal hypoglycemia risk and no difference in severe hypoglycemia compared with soluble human insulin.²¹ Those concerns and advantages should be taken into account when patients, especially those who have long interval between meals, are about to switch from human insulin to insulin aspart.

To counteract the slower absorption, human insulin was injected 30 min before the meal in this study. Unfortunately, in daily life, the majority of patients (75%) with type 1 diabetes fail to follow this recommendation.²² Because insulin aspart preparations can be injected immediately before the meal, they may offer more flexibility, compliance, and convenience than human insulin.

This study had certain limitations that should be taken into consideration when the results were interpreted. We had to choose an open-labeled trial design because (1) these insulins were easily distinguishable by visual inspection and (2) injection times of insulin aspart and human insulin were different. This design could have induced more biases on the evaluation of safety end points than a double-blinded study. In order to maintain optimal baseline glucose levels in the night before the study days, insulin aspart was infused intravenously before human insulin injection, and human insulin was infused before either of aspart preparations. This switch intended to minimize overlap on serum insulin concentration from intravenous infusions and subcutaneous injections. Baseline glucose levels were adjusted in the statistical analysis, and means of fasting plasma glucose were roughly equal among treatments. Furthermore, this was an experimental study with crossover trial design. We referred to an early study and calculated insulin dose based on patients' body weight rather than meal size in order to mimic a standard clinical condition and reduce the potential of hypoglycemic episodes. Thus, for some insulin preparations, this dose might not be effective for controlling postprandial glucose excursions.

In conclusion, this study suggests that BIAsp70 and BIAsp50 injected immediately before the meal are at least as effective as human insulin injected 30 min earlier. However, BIAsp50 showed the greatest similarity to human insulin in the pharmacokinetic and pharmacodynamic comparisons during our 12-h study period. If mealtime injection of human insulin is about to be replaced by an analog preparation with similar pharmacokinetic and pharmacodynamic properties, the most logical choice is probably BIAsp50. However, long-term study in the clinical situation is needed to confirm our observations gained from the current pharmacokinetic and pharmacodynamic study.

Footnotes

Acknowledgments

We would like to thank our research nurses Karen Lund and Ulla Sigsgaard for their efforts in dealing with practical issues. The collaborations from the Good Clinical Practice unit in Aarhus University Hospital and Assay Technology, Novo Nordisk A/S, are also highly appreciated. The study was supported by an unrestricted grant from Novo Nordisk A/S.

Author Disclosure Statement

Z.M. is the recipient of unrestricted grants for research from Novo Nordisk. T.L. holds shares in Novo Nordisk and has received unrestricted funding from Novo Nordisk A/S, Novo Nordisk Scandinavia, Astra Denmark, Pfizer Denmark, GlaxoSmithKline Pharma Denmark, Servier Denmark, and HemoCue Denmark. J.S.C. has received lecture and consultant fees and is the recipient of unrestricted research grants from Novo Nordisk. All other authors have nothing to declare.

References

American Diabetes Association. Insulin administration. Diabetes Care, 1998; 21,Suppl 1:572–575.

Home

, Alberti

. Human insulin. Clin Endocrinol Metab, 1982; 11:453–483.

Binder

, Lauritzen

, Faber

, Pramming

. Insulin pharmacokinetics. Diabetes Care, 1984; 7:188–199.

Ciofetta

, Lalli

, Del

, Torlone

, Pampanelli

, Mauro

, Chiara

, Brunetti

, Bolli

. Contribution of postprandial versus interprandial blood glucose to HbA1c in type 1 diabetes on physiologic intensive therapy with lispro insulin at mealtime. Diabetes Care, 1999; 22:795–800.

Lindholm

, Jacobsen

. Clinical pharmacokinetics and pharmacodynamics of insulin aspart. Clin Pharmacokinet, 2001; 40:641–659.

Lindholm

, McEwen

, Riis

. Improved postprandial glycemic control with insulin aspart. A randomized double-blind cross-over trial in type 1 diabetes. Diabetes Care, 1999; 22:801–805.

Bretzel

, Arnolds

, Medding

, Linn

. A direct efficacy and safety comparison of insulin aspart, human soluble insulin, and human premix insulin (70/30) in patients with type 2 diabetes. Diabetes Care, 2004; 27:1023–1027.

Kang

, Creagh

, Peters

, Brange

, Volund

, Owens

. Comparison of subcutaneous soluble human insulin and insulin analogues (AspB9, GluB27; AspB10; AspB28) on meal-related plasma glucose excursions in type I diabetic subjects. Diabetes Care, 1991; 14:571–577.

Rosenfalck

, Thorsby

, Kjems

, Birkeland

, Dejgaard

, Hanssen

, Madsbad

. Improved postprandial glycaemic control with insulin aspart in type 2 diabetic patients treated with insulin. Acta Diabetol, 2000; 37:41–46.

10.

Home

, Lindholm

, Hylleberg

, Round

. Improved glycemic control with insulin aspart: a multicenter randomized double-blind crossover trial in type 1 diabetic patients. UK Insulin Aspart Study Group. Diabetes Care, 1998; 21:1904–1909.

11.

Kaku

, Matsuda

, Urae

, Irie

. Pharmacokinetics and pharmacodynamics of insulin aspart, a rapid-acting analog of human insulin, in healthy Japanese volunteers. Diabetes Res Clin Pract, 2000; 49:119–126.

12.

Hirsch

, Farkas-Hirsch

, Skyler

. Intensive insulin therapy for treatment of type I diabetes. Diabetes Care, 1990; 13:1265–1283.

13.

Tamas

, Marre

, Astorga

, Dedov

, Jacobsen

, Lindholm

. Glycaemic control in type 1 diabetic patients using optimised insulin aspart or human insulin in a randomised multinational study. Diabetes Res Clin Pract, 2001; 54:105–114.

14.

Devries

, Lindholm

, Jacobsen

, Heine

, Home

. A randomized trial of insulin aspart with intensified basal NPH insulin supplementation in people with Type 1 diabetes. Diabet Med, 2003; 20:312–318.

15.

Heise

, Eckers

, Kanc

, Nielsen

, Nosek

. The pharmacokinetic and pharmacodynamic properties of different formulations of biphasic insulin aspart: a randomized, glucose clamp, crossover study. Diabetes Technol Ther, 2008; 10:479–485.

16.

Thorisdottir

, Parkner

, Chen

, Ejskjaer

, Christiansen

. A comparison of pharmacokinetics and pharmacodynamics of biphasic insulin aspart 30, 50, 70 and pure insulin aspart: a randomized, quadruple crossover study. Basic Clin Pharmacol Toxicol, 2009; 104:216–221.

17.

Jacobsen

, Sogaard

, Riis

. Pharmacokinetics and pharmacodynamics of a premixed formulation of soluble and protamine-retarded insulin aspart. Eur J Clin Pharmacol, 2000; 56:399–403.

18.

Raja-Khan

, Warehime

, Gabbay

. Review of biphasic insulin aspart in the treatment of type 1 and 2 diabetes. Vasc Health Risk Manag, 2007; 3:919–935.

19.

Mixtard 30—going, going, gone?. Drug Ther Bull, 2010; 48:85.

20.

Home

, Hallgren

, Usadel

, Sane

, Faber

, Grill

, Friberg

. Pre-meal insulin aspart compared with pre-meal soluble human insulin in type 1 diabetes. Diabetes Res Clin Pract, 2006; 71:131–139.

21.

Rys

, Pankiewicz

, Lach

, Kwaskowski

, Skrzekowska-Baran

, Malecki

. Efficacy and safety comparison of rapid-acting insulin aspart and regular human insulin in the treatment of type 1 and type 2 diabetes mellitus: a systematic review. Diabetes Metab, 2011; 37:190–200.

22.

Overmann

, Heinemann

. Injection-meal interval: recommendations of diabetologists and how patients handle it. Diabetes Res Clin Pract, 1999; 43:137–142.