A Double-Blind, Placebo-Controlled Trial Assessing Pramlintide Treatment in the Setting of Intensive Insulin Therapy in Type 1 Diabetes

  1. Steve Edelman, MD1,
  2. Satish Garg, MD2,
  3. Juan Frias, MD3,
  4. David Maggs, MD3,
  5. Yan Wang, PHD3,
  6. Bei Zhang, MD, MS3,
  7. Susan Strobel, PHD3,
  8. Karen Lutz, PHD3 and
  9. Orville Kolterman, MD3
  1. 1Division of Diabetes/Metabolism, San Diego VA Medical Center, San Diego, California
  2. 2Barbara Davis Center for Childhood Diabetes, University of Colorado Health Sciences Center, Denver, Colorado
  3. 3Amylin Pharmaceuticals, San Diego, California
  1. Address correspondence and reprint requests to Orville Kolterman, MD, Senior Vice President Clinical/Regulatory Affairs, Amylin Pharmaceuticals, 9360 Towne Centre Dr., San Diego, CA 92121. E-mail: okolterman{at}


OBJECTIVE—To assess safety, efficacy, and tolerability of pramlintide dose escalation with proactive mealtime insulin reduction, followed by insulin optimization, in patients with type 1 diabetes.

RESEARCH DESIGN AND METHODS—This 29-week, double-blind, placebo-controlled study randomized 296 patients to pramlintide or placebo as an adjunct to insulin. During initiation, pramlintide was escalated from 15 to 60 μg/meal (15-μg increments) with recommended reductions (30–50%) in mealtime insulin. Insulin was subsequently adjusted to optimize glycemic control. End points included safety and change in HbA1c (A1C), postprandial glucose, insulin, weight, and tolerability.

RESULTS—Baseline A1C was 8.1% for both groups and at week 29 had decreased comparably (pramlintide −0.5% [95% CI −0.61 to −0.33]; placebo −0.5% [−0.63 to −0.35]). Pramlintide treatment significantly reduced postprandial glucose excursions (incremental area under the curve [AUC]0–3h: pramlintide −175 ± 40, placebo −64 ± 38 mg · h−1 · dl−1; P < 0.0005) and weight (pramlintide −1.3 ± 0.30, placebo +1.2 ± 0.30 kg; P < 0.0001). At week 29, insulin dose decreased by 28 and 4% in pramlintide- and placebo-treated groups, respectively. Nausea, reported by 63 and 36% of patients in pramlintide and placebo groups (P < 0.01), respectively, was predominately mild to moderate in intensity. Severe hypoglycemia rates were low in both groups (pramlintide 0.57 ± 0.09, placebo 0.30 ± 0.06 event rate/patient-year; P < 0.05), with increased rates observed in patients remaining at 30 μg pramlintide.

CONCLUSIONS—Pramlintide dose escalation with reduced mealtime insulin was effective during therapy initiation in patients with type 1 diabetes. While both groups experienced equivalent A1C reductions relative to placebo, pramlintide-treated patients experienced reductions in postprandial glucose excursions and weight, not achievable with insulin therapy alone.

Amylin, a glucoregulatory hormone cosecreted with insulin from pancreatic β-cells (1), regulates postprandial glucose appearance by slowing gastric emptying (2), suppressing postprandial glucagon secretion (3), and reducing food intake (4,5). These actions complement effects of insulin; thus, both hormones act together to maintain postprandial glucose homeostasis. Amylin is deficient in patients with type 1 diabetes and in insulin-using patients with type 2 diabetes (1).

Pramlintide, the active ingredient in the SYMLIN (pramlintide acetate) injection (6), is an analog of amylin and the first noninsulin treatment for patients with type 1 diabetes since insulin was introduced. It is indicated as adjunct treatment in patients with type 1 and type 2 diabetes who use mealtime insulin and have not achieved desired glycemic control despite optimal insulin therapy. Through actions similar to those of amylin, pramlintide reduces postprandial glucose concentrations, improves overall glycemic control (HbA1c [A1C]) (7,8), and results in significant weight reductions compared with insulin therapy alone (9).

In previous studies, pramlintide-treated patients with type 1 diabetes experienced more nausea and severe hypoglycemia during therapy initiation than placebo-treated patients (7,8). Nausea was dose dependent, mostly mild or moderate in intensity, occurred early in therapy, and dissipated over time (7,8). Risk of insulin-induced severe hypoglycemia was increased, especially during the initial 4 weeks of therapy. Pramlintide was initiated at a fixed dose (no titration), and mealtime insulin dose was not reduced upon pramlintide initiation.

This study investigates whether progressive pramlintide dose escalation with concomitant reduction of mealtime insulin increases tolerability (reduces nausea) and reduces risk of severe hypoglycemia upon pramlintide initiation in patients with type 1 diabetes and whether, with subsequent optimization of insulin therapy, combining pramlintide and insulin offers unique metabolic effects not achieved with insulin alone.


A total of 296 patients with type 1 diabetes using intensive insulin therapy (multiple daily injections [MDIs] or continuous subcutaneous insulin infusion [CSII]) and self-monitoring blood glucose three or more times daily were enrolled. Inclusion criteria included age ≥18 years, insulin use >1 year, A1C 7.5–9.0%, and no severe hypoglycemia for 6 months before screening. Female subjects were postmenopausal, surgically sterile, or using contraception. Exclusion criteria included clinically significant comorbid conditions including gastroparesis, using medications affecting gastrointestinal motility, or using oral antidiabetic or antiobesity agents.

An institutional review board approved the protocol, and all patients provided written informed consent before participation. The study was conducted in accordance with principles described in the Declaration of Helsinki (1964), including all amendments through the South Africa revision (1996).

This 29-week, double-blind, placebo-controlled, multicenter study (29 U.S. centers) randomized patients to pramlintide or placebo (Amylin Pharmaceuticals, San Diego, CA) for a 4-week initiation period followed by a 25-week maintenance period. Study medication was administered subcutaneously immediately before meals, 3–4 times daily, depending on individual meal patterns. Insulin was administered separately.

During initiation, pramlintide was introduced at 15 μg/meal and increased weekly (15-μg increments) to 60 μg/meal (maintenance dose). Patients unable to achieve the 60-μg dose were treated with 30 μg/meal for the remainder of the study (one patient each remained at doses of 15 and 45 μg). A 30–50% reduction in mealtime insulin was recommended at pramlintide initiation to reduce risk of hypoglycemia. Subsequently, patients were instructed to reduce or increase basal insulin if preprandial glucose was <7.2 or >9.9 mmol/l, respectively, and reduce or increase mealtime insulin if postprandial glucose was <8.8 or >13.3 mmol/l, respectively.

During the maintenance period, pramlintide dose was held constant, while insulin dose was adjusted to achieve prespecified glycemic targets (preprandial: 6.1–7.7 mmol/l; postprandial: 7.7–9.9 mmol/l) based on self-monitored blood glucose. Insulin dosing adjustments for the placebo-treated group followed the same guidelines outlined for the pramlintide-treated group.

Patients were instructed to self-monitor blood glucose concentrations six or more times daily throughout the study (including measurements before and 1–2 h after meals). Glucose concentrations, insulin doses, and hypoglycemia symptoms were recorded in electronic diaries (INVIVOSYSTEM; invivodata, Scotts Valley, CA). Diary data were uploaded daily to a secure internet-based server allowing assistance in insulin adjustments by study personnel. Patients were evaluated weekly during initiation and monthly during maintenance. Diary entries, body weight, and adverse events were assessed at each visit. A1C was measured at baseline and monthly thereafter.

Seventy-seven patients underwent standardized meal tests (30% total daily calories; 55% carbohydrate, 15% protein, and 30% fat) on day 1 (baseline) and at weeks 4, 16, and 29. Plasma glucose samples were collected immediately before and at various intervals for 3 h postmeal. Plasma glucose concentrations were measured by a central laboratory (Quintiles Laboratories, Smyrna, GA).

Safety evaluations were based on reports of treatment-emergent adverse events, laboratory evaluations, and physical examinations. The intensity of each adverse event was defined as mild (transient, not interfering with daily activities), moderate (low level of inconvenience, intermittently interfering with daily activities), or severe (interrupting daily activities).

Hypoglycemia was categorized as nonsevere (mild or moderate) or severe. Severe hypoglycemia was defined using Diabetes Control and Complications Trial (10) criteria. The assistance of another person was required to obtain treatment for the event, including administration of intravenous glucose or intramuscular glucagon. Symptoms consistent with hypoglycemia or a self-monitored blood glucose concentration <60 mg/dl, regardless of presence of symptoms, not meeting the criteria for severe hypoglycemia, were categorized as nonsevere hypoglycemia.

End points

Primary end points were safety assessments (including severe hypoglycemia). Secondary end points included change from baseline to various time points in A1C, postprandial glucose concentrations, insulin doses, and body weight.

Statistical analyses

A minimum of 125 patients per treatment group provided reliable safety data and ∼88% power for testing noninferiority of pramlintide (plus insulin) to placebo (plus insulin) relative to A1C change from baseline to week 29 using a 0.4% margin. Power calculations assumed mean A1C change was −0.5% with an SD of 1.0% for both treatment groups.

Study end points, baseline demographics, and subject disposition were descriptively summarized by treatment group and by maintenance dose within the pramlintide group for the intent-to-treat population. Treatment-emergent adverse events were presented as subject incidence. Severe hypoglycemic episodes were presented as annual event rate per subject year of treatment (SE) and were calculated for the entire study course and by study period (initiation and maintenance). Although the study was not powered to detect differences in safety outcomes, data for the adverse events of note were analyzed parametrically using Fisher’s exact test and Z test.

For efficacy end points, mean ± SE and median values were calculated at each visit. A1C change from baseline to week 29 was parametrically analyzed, within and between treatment groups, using a general linear model. Model covariates included treatment (pramlintide/placebo), study site, and baseline A1C. Least squares (LS) mean, SE, and 95% CIs for LS mean were derived for each treatment. For the between-group analysis, the upper one-sided 95% CI was calculated for LS mean difference. Noninferiority was concluded if the upper one-sided 95% CI was <0.4%. All statistical tests for the within-group analysis were two sided using a significance level of 0.05 (11). Similar analysis was performed for change in incremental plasma glucose concentration area under the curve (AUC)0–3h. Parametrical analysis of weight change between treatment groups was performed using a general linear model and a significance level of 0.05.


Baseline demographics and patient disposition

Baseline demographics were well matched between treatment groups. Patients used MDIs or CSII, and mean baseline A1C in both groups was ∼8% (Table 1).

Of 296 patients randomized, 117 (79%) pramlintide-treated and 133 (90%) placebo-treated patients completed the study. Most withdrawals were due to withdrawal of consent (pramlintide 8.1%, placebo 3.4%) or adverse events (pramlintide 5.4%, placebo 2.0%). Of pramlintide-treated patients completing the study, 78% escalated to the 60-μg dose, 21% remained at the 30-μg dose, and ∼1% remained on the 15- and 45-μg doses.

Glycemic control


As per study design, mean change in A1C from baseline to week 29 was comparable for pramlintide (−0.5% [95% CI −0.61 to −0.33]) and placebo (−0.5% [−0.63 to −0.35]) groups (both P < 0.0001 compared with baseline). The 95% upper CI for difference in change in A1C between pramlintide and placebo was 0.19% (less than the 0.4% noninferiority margin).

Postprandial glucose.

At baseline, postprandial glucose profiles from the meal test were similar between pramlintide and placebo groups (Fig. 1A and B). After 29 weeks of therapy, mean incremental plasma glucose AUC0–3h (±SE) was significantly reduced compared with baseline in the pramlintide (−175 ± 40 mg · h−1 · dl−1, P = 0.0002) but not the placebo (−64 ± 38 mg · h−1 · dl−1) group (Fig. 1A and B).

Self-monitored blood glucose concentrations revealed that postprandial glucose reductions were sustained over 29 weeks in pramlintide-treated patients. Throughout the study, a greater percentage of pramlintide-treated patients achieved postprandial glucose concentrations below the American Diabetes Association’s recommended target of 9.9 mmol/l at breakfast (68%), lunch (71%), and dinner (70%) compared with placebo-treated patients at breakfast (51%), lunch (61%), and dinner (58%) (P < 0.0001 for each meal).

Insulin dose

Total daily insulin doses following 29 weeks of therapy were reduced by ∼12% in pramlintide-treated patients with little change (+1%) in placebo-treated patients. During the initial 4 weeks of therapy, mealtime insulin dose decreased by ∼28 and ∼8% in pramlintide- and placebo-treated patients, respectively. Throughout the remainder of the study, mealtime insulin dose remained unchanged (pramlintide) and was reduced by ∼4% (placebo) compared with baseline (Fig. 2A). Basal insulin doses for pramlintide- and placebo-treated patients were unchanged during the 4-week initiation period and subsequently increased by ∼3 and 10%, respectively.


Pramlintide-treated patients achieved significant (P < 0.0001) reductions in mean (±SE) body weight versus placebo (change from baseline to week 29: pramlintide −1.3 ± 0.30 kg and placebo +1.2 ± 0.24 kg) (Fig. 2B).

Safety and tolerability

Overall, adverse events were of mild to moderate intensity. Adverse events, other than nausea and hypoglycemia, occurring with an incidence >10% in any treatment group and with a twofold-greater incidence in any pramlintide treatment group versus placebo included reduced appetite, vomiting, and sinusitis (Table 2). Nausea was predominantly mild to moderate, with <5% of pramlintide-treated patients and <1% of placebo-treated patients reporting severe nausea (Table 2). Consistent with study design, nausea was observed less frequently in those patients titrating to the 60-μg pramlintide dose than in patients remaining at the 30-μg pramlintide dose.

Overall, both treatment groups experienced comparable incidence of nonsevere hypoglycemia symptoms (pramlintide 92% and placebo 91%) and glucose concentrations <60 mg/dl (pramlintide 95% and placebo 94%), suggesting that both groups recognized hypoglycemic symptoms and comparably experienced nonsevere hypoglycemia. Event rates of severe hypoglycemia during initiation were similar in all groups except the 30-μg pramlintide group (Table 2). During maintenance, event rates of severe hypoglycemia for pramlintide-treated patients decreased, except in those patients in the 30-μg group (Table 2). Five patients withdrew due to adverse events (mild to moderate nausea [two patients, 15-μg dose], severe hypoglycemia [one patient, 30-μg dose], and depression [one patient, 30-μg dose]). One placebo-treated patient withdrew due to nausea.


This study demonstrated an effective clinical approach for using pramlintide as an adjunct to mealtime insulin in intensively treated patients with type 1 diabetes. Pramlintide dose escalation with a concomitant reduction in mealtime insulin, followed by insulin dose optimization, led to an improved safety profile compared with previous clinical trials (7,8) not using this initiation regimen while continuing to demonstrate improved glycemic and weight parameters.

As with previous clinical trials, the most common adverse events associated with pramlintide were gastrointestinal in a nature. Approximately 63 and 36% of pramlintide- and placebo-treated patients, respectively, experienced nausea at some time during the course of therapy. Though nausea was more common in pramlintide-treated patients, it was primarily mild to moderate in intensity and tended to dissipate with time. Most patients (∼70%) were able to titrate to the 60-μg dose, and only two pramlintide-treated patients (∼1.4%) withdrew due to nausea. In previous clinical trials, withdrawal due to nausea was ∼9%. Improved tolerability in this study was likely due to initial pramlintide dose escalation from 15 μg to the target dose of 60 μg per meal.

In previous studies in which pramlintide was initiated at a fixed dose and mealtime insulin was not proactively reduced, the event rate per patient-year of severe hypoglycemia during the initial 4 weeks of therapy ranged from ∼2.0 to 4.0 (7,8). In the present study, where mealtime insulin dose was proactively reduced by 30–50% upon pramlintide initiation, the rate of severe hypoglycemia during the initial 4 weeks of therapy was 0.75 events/patient-year. Most episodes of severe hypoglycemia occurred in patients remaining at 30 μg due to tolerability issues related to nausea. In clinical practice, these patients should be monitored more closely for potential occurrence of severe hypoglycemia. Though nausea did not significantly affect most pramlintide-treated patients, clinicians and patients must be aware of the potential for pramlintide-induced nausea and make appropriate adjustments in mealtime insulin to mitigate the risk of hypoglycemia.

As anticipated, given a study design in which both groups adjusted insulin doses to achieve similar glycemic targets required for the evaluation of the pramlintide safety profile, A1C reduction in both groups was similar. Pramlintide-treated patients, however, achieved this A1C with significant reductions in postprandial glucose concentrations, mealtime insulin dose, and weight.

Pramlintide-treated patients experienced significant reductions in glucose concentrations 1–2 h after meals. A1C reduction was similar in both groups, likely due to late postprandial waning of glucose control in pramlintide-treated patients (Fig. 1). Significant use of rapid-acting analogs such as mealtime insulin with minimal changes in basal insulin doses, coupled with pramlintide’s mechanisms of action, including slowing of gastric emptying, likely resulted in suboptimal insulin “coverage” in this late postprandial period. This finding is consistent with a study evaluating acute postprandial glucose-lowering effects of a single pramlintide injection in conjunction with either insulin lispro or regular insulin (12). In that study, rising glucose concentrations in the late postprandial period were observed with insulin lispro but not regular insulin. Therefore, several insulin-dosing strategies, including end-of-meal dosing of rapid-acting insulin analogs, use of regular human insulin as mealtime insulin, or use of insulin pump functions (“square wave” or “dual wave”) to extend bolus delivery, may lead to improved glucose control throughout the interprandial period and further reductions in A1C in pramlintide-treated patients.

Postprandial glucose control is clinically significant for two reasons. First, it is an important component of overall glycemic exposure (A1C). Monnier et al. (13) demonstrated that at A1C levels approaching the American Diabetes Association target of <7.0%, the relative contribution of postprandial glucose to overall glycemic control is ∼70%. Thus, agents specifically targeting postprandial hyperglycemia may be important in patients approaching, yet not achieving, A1C goals. Second, independent of its effect on the A1C, postprandial hyperglycemia has been implicated in development of micro- and macrovascular complications (1419). Hirsch and Brownlee (20) postulated that A1C alone may be an incomplete measure of glucose control and glucose fluctuations should also be considered when assessing the risk of long-term complications. Their hypothesis is based on Diabetes Control and Complications Trial data demonstrating that at similar A1C levels, conventionally treated patients had increased risk of retinopathy compared with intensively treated patients. This may have been due to reduced glucose fluctuations in intensively treated patients treated with mealtime insulin. Another hypothesis, explaining the relationship between postprandial hyperglycemia and diabetes complications, centers on vascular damage caused by hyperglycemia-induced oxidative stress (21).

Despite significant improvements in A1C and postprandial glucose, mean reductions in weight in pramlintide-treated patients after 29 weeks of therapy was 1.3 kg. In contrast, placebo-treated patients (insulin alone) gained an average of 1.2 kg. Pramlintide-induced weight reduction is thought to arise from its central effect on satiety, resulting in reduced food intake (9).

Weight gain is an important clinical barrier to optimizing glycemic control (2225). It is a frequent cosmetic deterrent to intensification of insulin therapy and has been associated with various components of the metabolic syndrome, even in patients with type 1 diabetes, potentially increasing long-term cardiovascular risk (23). Evidence that weight gain may play a role in significantly increased risk of coronary artery disease in type 1 diabetes is supported by the observation that risk of progression of coronary artery calcification increases with higher insulin doses in overweight patients with type 1 diabetes (26).

Dose escalation combined with proactive reduction in mealtime insulin provides an effective method for introducing pramlintide into the treatment regimen of intensively treated patients with type 1 diabetes. This method of pramlintide initiation, followed by insulin dose optimization, resulted in a combination of metabolic effects not achievable with insulin alone.

Figure 1—

Postprandial glucose concentrations after a meal test (n = 77) administered at weeks 0 and 29 in placebo-treated (n = 44) (A) and pramlintide-treated (n = 33) (B) patients. Reduction in postprandial glucose measured between weeks 0 and 29 was significant (P = 0.0002, incremental AUC0–3h) in pramlintide-treated patients. ○, week 0; ▪, week 29.

Figure 2—

The effect of pramlintide on percent reduction (SE) in mealtime daily insulin dose (median) (A) and body weight (B) from weeks 0 to 29 compared with placebo in the intent-to-treat, last-observation-carried-forward population (n = 295). Reduction in body weight between weeks 0 and 29 was significant (P < 0.0001) in pramlintide-treated patients compared with placebo-treated patients. ○, placebo; ▪, pramlintide.

Table 1—

Baseline demographics and patient disposition

Table 2—

Adverse events with an incidence ≥10% in any group and twofold greater in any pramlintide- than placebo-treated group


  • A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances.

    The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

    • Accepted June 22, 2006.
    • Received January 10, 2006.


| Table of Contents