Pramlintide Reduces Postprandial Glucose Excursions When Added to Regular Insulin or Insulin Lispro in Subjects With Type 1 Diabetes: A dose-timing study

A total of 21 subjects with type 1 diabetes using insulin lispro and 19 subjects with type 1 diabetes using regular insulin underwent a standardized mixed meal test on 5 consecutive days. Subjects were 18–65 years of age and had the following characteristics: a history of type 1 diabetes for at least 1 year, a baseline A1C value of 7–11%, free from symptoms of severe hypoglycemia or severe hyperglycemia (defined according to Diabetes Control and Complications Trial criteria [24]) for 2 months, stable weight for 2 months, stable daily insulin dose (within ± 10%), and no change in type of insulin used for 2 months before the study. Women who were not surgically sterile or postmenopausal were to practice appropriate contraception. Subjects were excluded if they had clinically significant comorbid conditions or received concomitant treatment with drugs known to affect gastrointestinal motility, including but not limited to erythromycin, metoclopramide, cisapride, cholestyramine, or colestipol.

The Institutional Review Board of each study site approved the protocol, and all subjects provided written informed consent before participation. This study was conducted in accordance with principles described in the Declaration of Helsinki (1964), including all amendments up to and including the South Africa revision (1996).

Within 14 days after screening, consenting subjects were admitted to the clinical research center for at least 6 days. Between screening and admission, subjects were asked to record their daily food intake, insulin regimen, and self-monitored blood glucose results in a diary. Upon admission, each subject underwent a mixed meal test on 5 consecutive days in a randomized, single-blind, placebo-controlled, five-way crossover design. The meal was a standardized breakfast, consisting of a bagel with margarine and jam, cheese, yogurt, milk, and orange juice. The size of the meal was calculated individually to provide 30% of a subject’s total daily caloric requirements with a macronutrient composition according to the American Diabetes Association nutritional recommendations (55%/15%/30% of kcal from carbohydrate/protein/fat, respectively). The size of the standardized breakfast meal was the same on each study day for each individual, and the meal was always consumed within 15 min.

On each day, subjects received one of five treatments (a subcutaneous injection of placebo at −15 min or 60 μg pramlintide [0.6 mg/ml, Amylin Pharmaceuticals, Inc., San Diego, CA] at −15, 0, +15, or +30 min relative to the standardized breakfast) according to a randomized sequence, after an overnight fast. Pramlintide or placebo was injected into the subcutaneous tissue of the anterior abdominal wall, on the opposite side from the insulin injection.

Subjects’ short-acting insulin dose was adjusted appropriately for both the content of the standardized meal and the anticipated effects of pramlintide based upon the individual subject’s history of their usual dietary intake and insulin use. Regular insulin was injected at −30 min and insulin lispro at 0 min relative to the standardized meal, which was consistent with the respective package insert directions. Efforts were made to keep the short-acting insulin dose constant at the time of each of the standardized breakfast meal challenges. Deviations from the predetermined short-acting insulin dose were allowed only for safety reasons (dose reduction to avoid postprandial hypoglycemia if the premeal glucose was near-normal) but not for the purpose of glycemic control (dose increase to improve postprandial hyperglycemia). The basal insulin regimens, which were comprised of continuous subcutaneous insulin infusion (n = 8) or subcutaneous injections of intermediate- or long-acting insulin (n = 32), were held constant throughout the study.

Main pharmacodynamic parameters included the incremental plasma glucose area under the concentration time curve (AUC) from 0 to 2 h (AUC_{0–2 h}), incremental AUC_{0–4 h}, and the incremental plasma glucose concentrations at specific sampling times. The mean ± SEM incremental plasma glucose concentration profiles were calculated and plotted by treatment and by study group. For each study group, the pharmacodynamic parameter data were summarized descriptively and were analyzed using mixed effect models. The mixed effect models included treatment, treatment sequence, and period as fixed effects and subject-within-sequence as random effects.

The P values for comparisons among the least square means of the incremental AUC_{0–2 h}, incremental AUC_{0–4 h}, and incremental glucose concentrations at various time points between dose timings were provided. Because the study was not powered to perform multiple comparisons, the P values were not adjusted for multiple comparisons.

Safety evaluations were based on reports of treatment-emergent adverse events in response to nondirected questioning, clinical laboratory evaluations (hematology, serum chemistry, urinalysis), vital signs (blood pressure and pulse rate), electrocardiograms, and physical examinations in all subjects.

Of the 40 subjects who were randomized, 38 were included in the evaluation population. Two subjects (one from the regular insulin and one from the insulin lispro group) were excluded from the evaluation population because they received only four of the five possible treatments. The regular insulin and the insulin lispro groups had similar demographic characteristics with regard to sex, race, age, BMI, and diabetes duration; however, the baseline A1C was somewhat lower in the insulin lispro compared with the regular insulin group (Table 1).

The mean premeal glucose concentrations were comparable across all 5 study days in both the regular insulin and insulin lispro groups (Table 2). In both the regular insulin and insulin lispro groups, pramlintide injections at all four time points lowered the postprandial glucose excursion compared with placebo (Table 2).

As illustrated in Fig. 1 and Table 2, pramlintide injections at −15 and 0 min prevented the initial rise in plasma glucose (significant difference in incremental plasma glucose at 30 and 45 min) compared with the +15 and +30 min pramlintide injections in both the regular insulin and insulin lispro groups. The type of insulin used had an effect on the resulting glucose pharmacodynamic profile (Fig. 1). Thus, in subjects using regular insulin and having pramlintide injections at −15 or 0 min, plasma glucose concentrations decreased slightly from baseline during the first hour, followed by a slight rise during the next 2 h before reaching a plateau (Fig. 1A). Comparatively, in subjects using insulin lispro and having pramlintide injections at −15 or 0 min, the decrease in plasma glucose concentrations was more pronounced during the first hour followed by a more pronounced rise during the subsequent 3 h (Fig. 1B).

As shown in Table 2, in both the regular insulin and insulin lispro groups, the AUC_{0–2 h} for all four pramlintide dose timings was significantly reduced compared with the placebo group (P < 0.0001 and P < 0.0026, respectively). In the regular insulin group, administration of pramlintide at 0 min elicited a significantly (P < 0.0001) greater reduction in AUC_{0–2 h} than the +15-min dose timing (Table 2). In the insulin lispro group, administration of pramlintide at −15 min elicited a significantly greater reduction in AUC_{0–2 h} than the +30-min dose timing, and administration of pramlintide at 0 min elicited a significantly greater reduction in AUC_{0–2 h} than the +15- or +30-min dose timings (Table 2).

In the regular insulin group, the AUC_{0–4 h} for all four pramlintide dose timings was significantly (P < 0.03) reduced compared with the placebo group (Table 2). In the insulin lispro group, only the AUC_{0–4 h} for the 0-min pramlintide injection was significantly (P < 0.0098) reduced compared with the placebo group (Table 2).

The mean preprandial doses of regular insulin and insulin lispro administered with the standardized test meal were comparable on each of the five meal test days (Table 2). Compared with day −1, the short-acting insulin dose administered at each of the five meal challenge tests was ∼30% reduced (Table 2).

There were no severe hypoglycemic episodes, no serious adverse events, and no clinically relevant changes in laboratory tests, vital signs, electrocardiograms, or abnormal findings upon physical examinations.

The overall incidence of treatment-emergent adverse events was similar among the regular insulin and insulin lispro groups with mild to moderate hypoglycemia and mild nausea being the most frequent treatment-emergent adverse events. In the regular insulin group, the incidence of hypoglycemia was 15.8, 26.3, 27.8, 21.1, and 21.1% and the incidence of nausea was 10.5, 21.1, 16.7, 10.5, and 26.3% in the placebo, −15-, 0-, +15-, and +30-min pramlintide dose arms, respectively. In the insulin lispro group, the incidence of hypoglycemia was 28.6, 28.6, 33.3, 28.6, and 20.0% and the incidence of nausea was 0, 4.8, 9.5, 14.3, and 5.0% in the placebo, −15-, 0-, +15-, and +30-min pramlintide dose arms, respectively. The vast majority of postprandial hypoglycemic events in the pramlintide dosing groups occurred when the fasting plasma glucose concentration was <7.0 mmol/l (9 of 10 episodes in the regular insulin group and 11 of 14 episodes in the insulin lispro group).

Several previous placebo-controlled studies in subjects with type 1 diabetes have consistently shown that preprandial administration of pramlintide reduced the early postprandial rise and overall postprandial excursion of plasma glucose when used in conjunction with regular insulin (17–19). However, these previous studies have not assessed the effects of pramlintide when given at various injection times or when given with a rapid-acting insulin analog. In this respect, the present study revealed several novel, clinically important findings.

The results from the insulin lispro group indicate that pramlintide also reduces the postprandial glucose excursion when used as an adjunct to a rapid-acting insulin analog. This is a clinically important finding, given that rapid-acting insulin analogs have themselves been shown to improve postprandial glucose excursion compared with regular insulin in patients with type 1 diabetes (25–27). In these previous studies (25–27), rapid-acting insulin analogs reduced the overall postprandial glucose excursion, but the initial rise in postprandial glucose was typically unaffected, compared with regular insulin. This may be explained by the fact that rapid-acting insulin analogs limit the postprandial glucose excursion in part by facilitating a more rapid uptake of glucose from the circulation into peripheral tissues. In contrast, pramlintide reduces the postprandial glucose excursion by tempering the appearance of both endogenous (liver-derived) and exogenous (meal-derived) glucose into the circulation (5–7).

A comparison between the regular insulin and insulin lispro groups showed that while pramlintide reduced the postprandial glucose excursion with both types of insulin, the profile of the postprandial glucose concentrations was influenced by, and in fact reflective of, the onset and duration of action of the concomitantly injected insulin. Thus, following mealtime injection of pramlintide as an adjunct to insulin lispro, the mean glucose concentration initially fell slightly (with a nadir at ∼60 min that coincided with the time at which insulin lispro reaches its peak action [26]), and then gradually rose as the action of insulin lispro tapered off. In contrast, following mealtime administration of pramlintide as an adjunct to regular insulin, the glucose concentration remained within a narrower range, consistent with the slower onset and longer duration of action of regular insulin.

The systematic evaluation of different dose-timing regimens showed that pramlintide administrations at either −15 min or immediately before a meal both prevented the early postprandial surge in plasma glucose that occurred with administration of regular insulin or insulin lispro alone (placebo). Administration of pramlintide at +15 or +30 min relative to the meal did not prevent the initial rise in plasma glucose but nevertheless still reduced the overall postprandial glucose excursion compared with administration of regular insulin or insulin lispro alone. This latter observation is entirely consistent with the mechanism of action of pramlintide: as the rate of glucose appearance is reduced to better match the rate of insulin-mediated glucose disappearance, even after the ingestion of the meal, the glycemic surge is curbed and plasma glucose concentrations revert toward lower levels.

The potent postprandial glucose lowering effect of pramlintide occurred despite a concomitant lowering of the mean preprandial short-acting insulin dose by ∼30%. For the purpose of the present study, it was important to keep both the dosing and the timing of insulin injections comparable across the five dose timings. However, it would be interesting in future studies to examine the postprandial glucose lowering effect of pramlintide in the context of various insulin dosing and timing regimens, such as in conjunction with postprandial insulin dosing, or in comparison with an increased preprandial insulin dose.

Although the study was not conducted in a clinical use setting and did not formally test different insulin dose adjustment regimens, it provides important information that may help guide appropriate insulin dose management when initiating pramlintide treatment in patients with type 1 diabetes. Subcutaneous administration of pramlintide 15 min before or immediately before meal ingestion is optimal for reducing postprandial glucose excursions, with differences in glucose excursion dependent on the duration of effect of the short- or rapid-acting insulin used. A reduction of the preprandial insulin dose (∼30–50%) is advised when initiating pramlintide treatment unless subjects persistently have preprandial blood glucose concentrations >14 mmol/l, in which case smaller reductions (or no reduction) may be appropriate. The finding that the majority of nonsevere hypoglycemic episodes during the postprandial period occurred when fasting glucose concentrations were <7 mmol/l indicates that even larger reductions in preprandial short-acting insulin doses may be required when the preprandial glucose concentration is in or near the normal range. In addition, increases in basal insulin coverage may be appropriate, particularly in subjects using rapid-acting insulin analogs, such as insulin lispro, for whom the importance of basal insulin adjustments has long been recognized (28). When used with appropriate adjustments to both basal and bolus insulin regimens, adjunctive treatment with pramlintide may lead to a reduction of excessive glucose fluctuations throughout the day, as recently documented by continuous glucose monitoring in subjects with type 1 diabetes treated intensively with insulin pumps (29).

We wish to thank Mark Fineman, Jonathan Kornstein, Terrie Burrell, and Mike Sierzega for their assistance in the conduct, reporting, and quality control of the study.