Objective. This study compared the glucose-lowering effect of insulin lispro, given before or after meals, with regular human insulin given before meals in prepubertal children with diabetes.
Research Design and Methods. A 3-way crossover, open-label study involving 61 prepubertal children (ages 2.9–11.4 years) with type 1 diabetes. The children were randomly assigned to receive regular human insulin 30 to 45 minutes before meals, insulin lispro within 15 minutes before or immediately after meals, combined with basal insulin. Each treatment lasted 3 months. Hemoglobin A1c levels and home glucose monitoring profiles were measured at the end of each treatment period.
Results. Treatment with insulin lispro before breakfast resulted in lower 2-hour postprandial glucose values than regular human insulin (11.7 ± 4.4 mmol/L vs 15.0 ± 5.4 mmol/L). Similarly, insulin lispro given before dinner resulted in lower blood glucose values 2 hours postprandially (8.8 ± 5.0 mmol/L vs 10.8 ± 5.4 mmol/L) than regular human insulin. When insulin lispro was administered after meals, the 2-hour glucose levels were between those seen with either insulin lispro or regular human insulin given before meals. The number and types of adverse events, the rates of hypoglycemia, and the HbA1c levels did not differ among the 3 therapies.
Conclusions. In prepubertal children, insulin lispro given before meals is safe and significantly lowers postprandial glucose levels after breakfast and dinner compared with regular human insulin, and insulin lispro given after the meal provides similar benefits as regular human insulin before the meal.
The aim of diabetes management is to prevent short- and long-term complications of the disease, while at the same time offering lifestyle flexibility. Multiple factors including activity level, emotions, and irregular meal schedules often make adequate glycemic control difficult to achieve in prepubertal children.1Thus, the treatment of diabetes in children should be directed toward achieving acceptable glucose control while minimizing the risk of hypoglycemia.
Regular human insulin injected subcutaneously typically results in serum insulin levels that peak later and remain elevated longer than those observed after pancreatic insulin secretion. As a result of the delayed onset and relative long duration of regular human insulin activity, patients may be at risk of hypoglycemia before the next meal. Insulin lispro, a rapid-acting analog of human insulin, was developed to address some of the disadvantages of regular human insulin. The change in the primary structure of insulin lispro, a reversal of the amino acid sequence at positions B28 and B29, decreases self-association between insulin molecules and allows faster absorption and shorter duration of activity after subcutaneous injection compared with regular human insulin.2 Because children tend to have unpredictable eating habits (eg, timing and size of meals), the pharmacokinetic properties of insulin lispro suggest that it may be especially useful in children with diabetes, even when it is given after meals. Although insulin lispro has been shown to lower postprandial glucose levels significantly compared with regular human insulin in adults,3,,4 less information is available in children,5,,6 particularly prepubertal children. In the current study, the safety and efficacy of insulin lispro was evaluated and compared with regular human insulin in prepubertal children.
RESEARCH DESIGN AND METHODS
This was a randomized, 3-period crossover, open-label study of 61 children with type 1 diabetes. The study was conducted at 6 investigative sites in the United States and Canada. Ten patients were studied at each of 5 centers; 11 children were studied at a sixth center. All patients had been diagnosed with diabetes for at least 12 months and had been receiving human insulin for at least the previous 2 months before randomization. Institutional review boards of each participating institution approved the protocol, and each guardian/parent gave written informed consent (children gave assent) for the study. According to the Declaration of Helsinki, all participants completed an informed consent form before entering the study.
During a 1- to 2-month lead-in period, the patients continued their basal insulin treatment with human NPH, lente, or ultralente, in addition to regular human insulin given before meals. After the lead-in period, patients were randomly assigned to 1 of 3 therapy sequence groups. Each treatment period lasted 3 months and consisted of 1 of the following: regular human insulin given 30 to 45 minutes before meals; insulin lispro given within 15 minutes before meals; or insulin lispro given directly after meals. With each treatment sequence, the patients also received basal insulin, which consisted of NPH, lente, or ultralente. The type of basal insulin and its injection frequency remained consistent throughout the study for all patients. At the end of each treatment period, the patient began the next treatment in sequence. Sequence A consisted of therapy with insulin lispro after meals, followed by regular human insulin before meals, then insulin lispro before meals. Sequence B consisted of insulin lispro before meals, followed by insulin lispro after meals, then regular human insulin before meals. Sequence C consisted of regular human insulin before meals, followed by insulin lispro before meals, then insulin lispro after meals.
Patients were seen monthly, with insulin doses adjusted as needed in an attempt to maintain glucose levels within individually defined target ranges.7 Because the study involved children 3 to 12 years of age, the same target was not feasible. In general, the target blood glucose range for younger children is kept higher than for older children. In addition, the target range for one child at a given age might differ from that of another child of the same age. Thus, individualized management is mandatory for this age range.7 This individual target range was maintained for the duration of the study, which allowed for intrapatient comparisons among treatments.
OneTouch blood glucose meters (Lifescan, Inc, Milpitas, CA) were provided by the investigator. Each patient/parent was instructed on use of the meter and their technique was assessed at each clinic visit. As described above, each physician was asked to set blood glucose targets for the study and code them in the software for downloading to the blood glucose meter provided. This target range was used in calculating the proportion of glucose values above, below, and within range. Software from Lifescan was used to download data from the meters and validate data from the diaries provided to each patient. Seven-point blood glucose profiles were conducted on 2 nonconsecutive days during the week before the randomization visit and the week before each treatment period ended. During those days, one of which was a weekend day, blood glucose measurements were obtained at 3am and directly before and 2 hours after the morning, noon, and evening meals. In addition, blood glucose data from the 2 weeks before clinic visits were downloaded from the meters and summary glucose data were recorded. Insulin doses, injection times, and times of meals and snacks were recorded in the patient diaries. For the purpose of this study, a snack was defined as the ingestion of food containing ≥15 g of carbohydrate. Because children had been diagnosed with diabetes for at least 1 year before entering the study, the dietary advice of the individual centers was used throughout the study. A dietician met with the patient/parent at baseline and during the study (visits 5 and 8) to ensure compliance with the recommended diet and assess need for modification based on any lifestyle changes.
Hemoglobin A1c, serum chemistries, and serum antibodies to human insulin and insulin lispro were measured at the randomization visit and at the end of each treatment period. HbA1c levels were determined using high performance liquid chromatography at a central laboratory (Covance Laboratories, Indianapolis, IN; reference range: 4.3%–6.1%). In addition, antibodies specific for human insulin and for insulin lispro, in addition to antibodies that cross-react with both insulins, were measured and expressed as percent binding.8 Hypoglycemic events were recorded in the patient diaries and collected at each visit. Hypoglycemia was defined as any time a patient had symptoms associated with hypoglycemia or had measured blood glucose <3.5 mmol/L. Severe hypoglycemia was defined as in the Diabetes Control and Complications Trial, as hypoglycemia needing the assistance of others, resulting in coma, or requiring intravenous glucose or glucagon. The hypoglycemia rate was expressed as the number of episodes adjusted for a 30-day period.
All treatment comparisons were performed using a 2-tailed test with a nominal significance level of 0.05. Data from all patients entered in this study were used in the analyses. However, only data from patients randomized to 1 of the 3 sequence groups were used in any treatment comparisons. Analyses were performed at each visit and at the endpoint, defined as the last measurement observed for each patient within each treatment period. The patient's baseline characteristics within each sequence group were compared using a 1-way analysis of variance or a χ2 test. The efficacy and safety variables measured using a numerical scale were analyzed using an analysis of variance model for a 3-period crossover study. This model included the effects of sequence group, period, and treatment. Pairwise treatment comparisons were performed using a t test with the least square means from PROC GLM in SAS software (SAS Institute, Inc, Cary, NC).9 No consistent carryover effects were observed in these analyses.
A total of 61 patients (29 males, 32 females) entered this study. Two patients discontinued the study before completion; 1 before randomization and 1 during treatment with regular human insulin (because of personal reasons). Thus, 59 children successfully completed this study. The average age of the patients was 7.6 years (range: 2.9–11.4 years), and the average duration of diabetes was 3.7 years (range: 1–8 years). At baseline, 12% of patients used lente as their basal insulin, 74% used NPH, and 14% used ultralente. The baseline HbA1c was 8.4 ± 1.0% (range: 5.8%–10.2%). Of the 60 randomized patients, 17 were randomized to Sequence A, 20 to Sequence B, and 23 to Sequence C. There were no statistical differences among the sequence groups with respect to patient characteristics at baseline.
The mean total daily insulin dose during each treatment was 0.8 ± 0.2 U/kg/d and did not differ significantly among the treatment groups (P > .05). The mean daily dose of insulin lispro when administered before meals was 0.21 U/kg/d, compared with regular human insulin (0.25 U/kg/d) and insulin lispro given after meals (0.23 U/kg/d). On average, 2 injections of basal insulin and fast-acting insulin (regular human insulin or insulin lispro) were given during the day (breakfast and dinner). The daily doses of basal insulin did not differ among the treatments (regular human insulin 0.58 units/kg, insulin lispro after meals 0.60 units/kg, insulin lispro before meals 0.61 units/kg; P = .189). Nine patients received regular insulin at the noon meal, while 10 patients received insulin lispro at noon. The reason that fewer injections were given at noon may relate to an inability to administer injections to young children in a day care center environment or at school.
Seven-Point Glucose Profile
The results of the 7-point glucose profiles from home blood glucose monitoring are shown in Fig 1. The glucose levels at 3 am and before breakfast were not significantly different among the 3 therapies. However, the 2-hour postprandial blood glucose levels after breakfast were significantly lower when insulin lispro was administered before the meal when compared with the other treatments. The mean 2-hour postprandial glucose levels after breakfast were 11.7 ± 4.4 mmol/L, 13.5 ± 5.5 mmol/L, 15.0 ± 5.4 mmol/L, for insulin lispro before breakfast, insulin lispro after breakfast, and regular human insulin before breakfast, respectively (P < .001). On the other hand, insulin lispro given after breakfast resulted in lower premeal glucose values at the noon meal (8.3 ± 3.1 mmol/L) compared with either insulin lispro given before breakfast (8.7 ± 3.9 mmol/L, P = nonsignificant [NS]) or regular human insulin (9.5 ± 4.1 mmol/L, P = .037). The 2-hour postprandial glucose levels after the noon meal did not differ among the 3 treatments, most likely because of the lack of an insulin injection at noon in the majority of patients.
No difference in glucose levels before the evening meal was observed among the therapies. However, similar to the pattern seen after breakfast, 2-hour postprandial glucose levels after dinner were lowest when insulin lispro was administered before the meal (8.8 ± 5.0 mmol/L) when compared with insulin lispro given after dinner (9.9 ± 4.7 mmol/L, P = NS) and regular human insulin before dinner (10.8 ± 5.4 mmol/L; P = .006).
The mean HbA1c at endpoint did not differ among the 3 therapies (insulin lispro before meals 8.40% ± 1.1%, regular human insulin before meals 8.43% ± 1.0%, insulin lispro after meals 8.54% ± 1.0%, P = NS). No correlation was observed between HbA1c and age.
Home Blood Glucose Monitoring
Data representing 2 weeks of glucose values at the end of each treatment period and before the clinic visits are shown in Table 1. On average, during the 2 weeks, blood glucose was measured 4 times each day. Neither the overall maximum glucose level nor the minimum glucose level observed differed among the therapies, and between 36% to 38% of glucose levels were within their target range, regardless of the therapy. The overall mean glucose level differed significantly among the therapies, with insulin lispro given before meals associated with the lowest mean glucose (P= .024). Although no difference was seen among therapies with respect to proportion of glucose values within target ranges, the proportion of glucose values above target ranges were significantly less during therapy with insulin lispro given before meals (P < .05) when compared with the other 2 treatments. Therapy with insulin lispro before meals was also associated with an increased proportion of glucose values below the target range, compared with insulin lispro given after meals (P < .05). However, no difference in the rate of hypoglycemia among any of the therapies was observed (see below).
The overall rate of hypoglycemia (episodes/30 days), including those events with only mild symptoms, did not differ significantly among the 3 therapies (insulin lispro after meals 13.6 ± 9.3, regular human insulin before meals 13.8 ± 9.8, insulin lispro before meals 14.7 ± 11.9, P > .05). Likewise, no significant difference in the number of hypoglycemic episodes by time-of-day was observed among the 3 therapies. Severe hypoglycemia occurred in 6 patients during therapy with regular human insulin, in 3 patients during therapy with insulin lispro after meals, and in 2 patients during therapy with insulin lispro before meals (P > .05). Results of antibody measurements revealed no consistent effect of any of the insulin treatments. With respect to other adverse events, no significant differences were observed among the 3 therapies. The most common adverse events were rhinitis, pharyngitis, cough, and fever, none of which showed a relationship to the insulin therapies.
The present study demonstrates that insulin lispro given shortly before the meal achieved better postprandial blood glucose levels after breakfast and dinner when compared with insulin lispro administered after the meal or regular human insulin given 30 to 45 minutes before the meal. Although not reaching statistical significance, the postprandial glucose levels associated with insulin lispro administered after meals were between those seen with either insulin lispro or regular human insulin given before meals. Administering insulin lispro after meals, especially in younger children, may allow parents to better match the insulin dose with the meal content. The slightly lower dose of insulin lispro given before meals, compared with regular human insulin, is consistent with other studies,5 suggesting that the rapid action of insulin lispro results in lower insulin requirements at meal time.
In toddlers with diabetes, Rutledge and colleagues5 first presented evidence that insulin lispro was effective when given after meals. In their study, postprandial administration of insulin lispro was compared with preprandial administration of regular human insulin and insulin lispro. The children receiving insulin lispro after meals had lower mean 2-hour glucose excursions when compared with those patients treated with preprandial regular human insulin (6.2 mmol/L vs 13.0 mmol/L, P = .04). However, there was no statistical difference in the mean 1-, 2-, and 4-hour glucose excursions between patients treated with postprandial insulin lispro compared with patients treated with preprandial insulin lispro. Rutledge and colleagues concluded that postprandial insulin lispro administration results in blood glucose values as good or better than those obtained with the same dose of regular human insulin given before the meal.5 The results of the current study are consistent with those reported by Rutledge et al.5
HbA1c values in the present study did not differ among the therapies, in contrast to some other studies involving insulin lispro use in children and adolescents.10 To achieve optimal glucose control while using insulin lispro, careful adjustment of the basal insulin component must be conducted.10 Lalli et al11 demonstrated that by increasing the frequency of NPH injection, along with prandial insulin lispro, HbA1c levels can be significantly reduced in type 1 diabetes, In the present study, basal insulin dosing frequency was kept consistent with that used during the lead-in period, so that the effects of the fast-acting insulins could be effectively compared.
Managing and preventing hypoglycemia is a difficult aspect of treatment of children with type 1 diabetes. Dietary inconsistencies and differences in activity levels contribute to the unpredictability of hypoglycemia in children, especially nocturnal hypoglycemia.12,,13 Porter and colleagues14reported an increased frequency of nocturnal hypoglycemia in young children being treated with twice daily NPH and regular insulin. Those investigators reported an association between nocturnal hypoglycemia and younger age patients (<5 years) and lowered glycosylated hemoglobin (HbA1c <8.5%). In contrast, in the present study there was no significant difference among treatments in the rate of hypoglycemia or the number of hypoglycemic episodes at different times of the day and night.
In a study of 2873 children from 18 countries, the rate of hypoglycemic episodes (resulting in seizures and coma) was higher in younger children (<8 years of age) and was also associated with lower hemoglobin A1c levels.15 However, results from a prospective study in 146 children indicated that near physiologic hemoglobin A1c levels can be obtained without increasing the risk of severe hypoglycemia (resulting in loss of consciousness) using a multiple-dose insulin therapy.16In a meta-analysis of several individual studies comparing insulin lispro with regular human insulin, Brunelle and colleagues17 showed that the incidence of severe hypoglycemia was significantly lower during use of insulin lispro. In the present study, severe hypoglycemia seldom occurred, and its low frequency precluded statistical analysis among the treatments. However, there was no increase in the overall rate of hypoglycemia, despite the marked improvements in postprandial glucose levels seen after breakfast and dinner with premeal insulin lispro administration.
Children with type 1 diabetes are faced with a life-long disease requiring an intense treatment regimen. Insulin lispro offers treatment flexibility because of convenience in the timing of injection. In the present study, lower postprandial blood glucose levels were observed after breakfast and dinner with insulin lispro given before meals compared with insulin lispro given after meals and regular human insulin given before meals, despite unchanged HbA1C.
Insulin lispro is well-tolerated, safe, and effective in prepubertal children.
We thank Casey Mast, PharmD, RPh, and Barbara N. Campaigne, PhD, for writing assistance, and Peggy Campbell for editorial assistance with this manuscript.
- Received December 18, 2000.
- Accepted June 21, 2001.
Reprint requests to (L.C.D.) Children's Clinic, 2416 East Plaza Dr, Tallahassee, FL 32308. E-mail:
- NS =
- ↵Brackenridge BP, Rubin RR. Sweet Kids: How to Balance Diabetes Control & Good Nutrition with Family Peace. Alexandria, VA: American Diabetes Association, Inc; 1996
- Anderson JH Jr.,
- Brunelle RL,
- Koivisto VA,
- et al.
- Rutledge KS,
- Chase HP,
- Klingensmith GJ,
- Walravens PA,
- Slover RH,
- Garg SK
- ↵Staged Diabetes Management Owner's Manual, Type 1. Minneapolis, MN: International Diabetes Center; 1995:20
- Fineberg NS,
- Fineberg SE,
- Anderson JH,
- Birkett MA,
- Gibson RG,
- Hufferd S
- ↵SAS Institute Inc. SAS/STAT User's Guide, Version 6, Volume 2.4th ed. Cary, NC: SAS Institute Inc; 1999
- Dorchy H,
- Roggemans MP,
- Willems D
- Lalli C,
- Cioffeta M,
- Del Sindaco P,
- et al.
- Ebeling P,
- Jansson P-A,
- Smith U,
- Lalli C,
- Bolli GB,
- Koivisto VA
- Mortensen HB,
- Hougaard P
- Nordfeldt S,
- Ludvigsson J
- Brunelle RL,
- Llewelyn J,
- Anderson JH Jr.,
- Gale EAM,
- Koivisto VA
- Copyright © 2001 American Academy of Pediatrics