Clinical Care/Education/Nutrition ORIGINAL ARTICLE Use of Insulin Aspart, a Fast-Acting Insulin Analog, as the Mealtime Insulin in the Management of Patients With Type 1 Diabetes PHILIP RASKIN, MD ANDERS RIIS, MSC more rapid subunit dissociation and sub- RICHARD A. GUTHRIE, MD LOIS JOVANOVIC, MD cutaneous absorption of the insulin (4–6). LAWRENCE LEITER, MD In IAsp, the proline at position B28 has been replaced by aspartic acid. Because of this modification, its onset of action occurs in ϳ10–20 min, with maximal serum con- centrations reached in ϳ45 min (7–12). OBJECTIVE — To compare long-term glycemic control and safety of using insulin aspart Previous clinical trials in healthy sub- (IAsp) with that of regular human insulin (HI). jects and in patients with type 1 diabetes have demonstrated that IAsp has a faster RESEARCH DESIGN AND METHODS — This was a multicenter randomized open- label 6-month study (882 subjects) with a 6-month extension period (714 subjects) that onset and shorter duration of action than enrolled subjects with type 1 diabetes. Subjects administered IAsp immediately before meals does HI (13–17). In a previous study in or regular HI 30 min before meals; basal NPH insulin was taken as a single bedtime dose in the patients with type 1 diabetes receiving IAsp as the mealtime insulin in a multiple-injec- majority of subjects. Glycemic control was assessed with HbA1c values and 8-point blood glu- cose profiles at 3-month intervals. tion regimen, lower postprandial blood glucose levels and improved overall RESULTS — Mean postprandial blood glucose levels (mg/dl ± SEM) were significantly lower glycemic control were observed in patients for subjects in the IAsp group compared with subjects in the HI group after breakfast (156 ± 3.4 treated with IAsp compared with those vs. 185 ± 4.7), lunch (137 ± 3.1 vs. 162 ± 4.1), and dinner (153 ± 3.1 vs. 168 ± 4.1), when assessed treated with HI (17). The improved after 6 months of treatment. Mean HbA values (% ± SEM) were slightly, but significantly, lower 1c glycemic control was not associated with an for the IAsp group (7.78% ± 0.03) than for the regular HI group (7.93% ± 0.05, P = 0.005) at increase in the incidence of hypoglycemic 6 months. Similar postprandial blood glucose and HbA1c values were observed at 12 months. Adverse events and overall hypoglycemic episodes were similar for both treatment groups. episodes in these patients. The present study was conducted to CONCLUSIONS — Postprandial glycemic control was significantly better with IAsp com- examine the long-term efficacy and safety pared with HI after 6 and 12 months of treatment. The improvement was not obtained at an of IAsp compared with HI as the mealtime increased risk of hypoglycemia. HbA1c was slightly, but significantly, lower for IAsp compared component of an intensive insulin regi- with HI at 6 and 12 months. men. This article reports the improvement in glycemic control achieved by patients Diabetes Care 23:583–588, 2000 using IAsp, compared with HI, with 6 and 12 months of treatment. apid-acting insulin analogs have been lispro, B10) differ from regular human developed to have a peak and dura- insulin (HI) by having changes in amino RESEARCH DESIGN AND Rtion of activity that more closely coin- acid sequence or composition in the B chain METHODS — This was a randomized cide with the postprandial blood glucose of the insulin molecule. These alterations open-label parallel-group study conducted peak (1–3). The rapid-acting insulin reduce the stability of the insulin mono- at 59 centers in the U.S. and Canada. Sub- analogs (e.g., insulin aspart [IAsp], insulin mer–monomer interaction, leading to a jects with type 1 diabetes were treated with IAsp or HI as part of an intensive insulin reg- imen for 6 months, and then could continue From the Southwestern Medical Center at Dallas (P.R.), University of Texas, Dallas, Texas; Mid-America Dia- their assigned treatment in a 6-month exten- betes Associates (R.A.G.), Wichita, Kansas; St. Michael’s Hospital Health Center (L.L.), Toronto, Ontario, sion of the study. The study was performed Canada; Novo Nordisk A/S (A.R.), Copenhagen, Denmark; and the Sansum Medical Research Institute (L.J.), in accordance with the Declaration of Santa Barbara, California. Helsinki and with the approval of local inde- Address correspondence and reprint requests to Philip Raskin, MD, University of Texas, Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd., Dallas, TX 75390-8858. pendent review boards. Written informed P.R., R.A.G., L.L., A.R., and L.J. received grant support from Novo Nordisk. L.L. serves as a member of consent was obtained from all subjects. the advisory board for Novo Nordisk Canada and received honoraria from Novo Nordisk. A.R. is employed by and holds stock in Novo Nordisk. P.R. and L.J. served as members on the Advisory Board for and received Subjects honoraria from Novo Nordisk. Abbreviations: ANCOVA, analysis of covariance; HI, regular human insulin; IAsp, insulin aspart. The study enrolled 882 men and women, A table elsewhere in this issue shows conventional and Système International (SI) units and conversion aged 18–75 years, who had type 1 diabetes factors for many substances. for at least 18 months. At baseline, subjects DIABETES CARE, VOLUME 23, NUMBER 5, MAY 2000 583 Insulin aspart and long-term glycemic control were to have a BMI Յ35.0 kg/m2 and an same targets of glycemic control as speci- cose value Ͻ45 mg/dl (2.5 mmol/l) or had Յ HbA1c level 11%. The exclusion criteria fied in the run-in period. classical symptoms of hypoglycemia (such included impaired hepatic, renal, or cardiac No lifestyle-adjusting advice (diet or as sweating, strong hunger, dizziness, and function, recurrent major hypoglycemia, physical activity) was given so that the tremor) and were able to deal with the active proliferative retinopathy, or a total study could be conducted to test the inher- episode on their own. A major hypo- daily insulin dose Ն1.4 IU/kg. Women ent capabilities of the study medications on glycemic event was one that the subject were excluded if they were pregnant, breast- a true cross-section of type 1 diabetic indi- could not treat by himself/herself or feeding, or not practicing contraception. viduals, without confounding the data required administration of parenteral glu- interpretation with the independent effects cose or glucagon. Treatment of lifestyle intervention. All eligible subjects underwent a 4- to Statistical analysis 5-week run-in period during which time Efficacy assessments Glycosylated hemoglobin data were ana- they were treated with a multiple-injection Efficacy was assessed using 8-point blood lyzed using analysis of covariance regimen consisting of HI (Novolin R; Novo glucose and HbA1c values. The 8-point (ANCOVA) with HbA1c at baseline as Nordisk, Bagsvaerd, Denmark) as a meal- blood glucose profiles were recorded in a covariate and treatment and center as fixed related insulin and NPH insulin (Novolin diary by the subject on a weekday during effects. The last observation carried for- N; Novo Nordisk) as the basal insulin. Dur- the last week of the run-in period and on a ward approach was used for missing data ing the run-in and trial periods, HI was weekday before the 6- and 12-month study in most analyses. Individual time points of administered 30 min before each meal and visits. The prandial blood glucose incre- the 8-point blood glucose profile and NPH insulin was administered at bedtime. ment was derived from the 8-point blood derived blood glucose end points and Subjects were instructed on the proper use glucose measurements by averaging the insulin dose measures were compared by of the One Touch II (LifeScan, Milpitas, differences between the blood glucose val- ANCOVA as above; all values at 6 and 12 CA) blood glucose meter to take self-mea- ues just before and 90 min after each of the months were corrected for baseline differ- sured blood glucose measurements for 3 meals (values before and after all 3 meals ences. Changes in insulin antibodies (to insulin dose adjustments. The dosages of were required for inclusion in the analysis). IAsp, HI, and cross-reacting) were com- both HI and NPH insulin were adjusted by The blood glucose average was calculated pared between treatments using a Student’s a sliding scale so that subjects might attain as the average of the means of the blood t test. The Mantel-Haenszel ␹2 test was the following glycemic targets: glucose measurements for each subject (at used to compare the number of patients in fasting/preprandial blood glucose level of least 6 values from the 8-point profile were each treatment group experiencing at least 90–144 mg/dl (5–8 mmol/l), postprandial required for inclusion in the analysis). 1 nocturnal (midnight to 6:00 A.M.) major blood glucose level (1–3 h after a meal) of The HbA1c level (assay range 3–18%) was hypoglycemic episode. Results are stated as Յ180 mg/dl (Յ10 mmol/l), and 2:00 A.M. determined using the Bio-Rad Diamat system means ± SEM adjusted for baseline values blood glucose level of 90–144 mg/dl (5–8 (Bio-Rad, Hercules, CA) on blood samples at and center effect, or as mean treatment dif- mmol/l). After 2 weeks, an additional break- baseline, 3, 6, 9, and 12 months postran- ference (95% CI), or as indicated. fast dose of NPH insulin could be added to domization (18,19). The daily meal-related the regimen if satisfactory control of the and basal insulin doses were also calculated.