Adult Studies: Procedure for Induction of Hypoglycemia

Experimentally induced hypoglycemia was achieved via insulin infusion and monitored via a second catheter for blood sampling. Procedures for induction of hypoglycemia were similar across the adult studies. The procedure in Study IGBC is described here and is depicted in Figure 4 for Study IGBJ. Each glucagon dosing visit was conducted after an overnight fast of at least 8 hours with a starting plasma glucose of greater than or equal to 5.1 mmol/L. If the starting plasma glucose level was greater than 11.1 mmol/L, a priming dose of 2 to 4 units of IV insulin may have been given. Hypoglycemia was induced by an IV infusion of regular insulin diluted in normal saline at a rate of 2 mU/kg/min. Once the plasma glucose level reached less than 5.1 mmol/L, the infusion rate may have been decreased at the investigator’s discretion to 1.5 or 1.0 mU/kg/min. The infusion rate may have been adjusted as necessary up to a rate of 3 mU/kg/min to reach the target nadir plasma glucose level of less than 2.7 mmol/L. During the insulin infusion to induce hypoglycemia, plasma glucose levels were measured no more than 10 minutes apart while the plasma glucose level was greater than 5.6 mmol/L and no more than 5 minutes apart when the plasma glucose level was less than 5.6 mmol/L.

Figure 4

▬▬▬▬▬. ▬▬▬▬▬

A blood sample was collected for glucagon level and glucose level analysis 5 minutes after the insulin infusion was stopped (immediately prior to glucagon administration, T = 0). The insulin level also was also measured. If the plasma glucose level reached less than 60 mg/dL after receiving insulin for at least 3 hours, the assigned glucagon for the visit was administered.

▬▬▬▬▬

Pediatric Study: Induction of Hypoglycemia

In the pediatric study (IGBB), the glucose targets were not as aggressive as they were in the adult studies. Each visit for a glucagon dose was conducted after an overnight fast of at least 8 hours. On arrival to the research centre, an IV catheter was inserted into an arm vein for blood sampling. For patients using an insulin pump for diabetes management, the basal insulin infusion rate was increased by 25 to 50% to cause a gradual decline in plasma glucose. Bolus doses of insulin equal to 1 hour of the patient’s basal rate and further increases in basal insulin rate were administered, as needed, to achieve the target glucose of less than 4.4 mmol/L. Patients on injection therapy received their usual dose of long-acting insulin analogue in the 24 hours prior to the visit. Insulin was administered at a rate of 1 mU/kg/min intravenously to reach the target glucose of less than 4.4 mmol/L. A priming dose of 2 to 4 units of insulin IV also was given if needed. For patients who arrived at the centre with a plasma glucose of less than 4.4 mmol/L, no additional insulin was administered, the randomized glucagon preparation was given immediately after IV access was obtained, and baseline blood samples were collected.

Inclusion and Exclusion Criteria

All three adult studies enrolled patients between the ages of 18 to 65 years with T1D. Study IGBC enrolled six patients with T2D but data for these patients were not summarized in this report because inferences from the T2D population would be highly uncertain and the sponsor did not perform analyses on this subgroup for all outcomes. Study IGBI enrolled exclusively patients with T1D. Study IGBJ enrolled approximately equal numbers of patients with T1D and T2D. All studies excluded patients who had experienced severe hypoglycemia during the month prior to the study start.

Baseline Characteristics

The median age of patients in the adult trials ranged from 31 years (interquartile range [IQR] = 21 to 41) in the North American study (IGBC) to 52 years (range = 21 to 70 years) in Study IGBJ. Most patients had longstanding diabetes with median duration since diagnosis of more than 11 years in the three adult studies (age range across adult studies: 1 to 43 years). Most patients in the North American study (IGBC) used an insulin pump as their primary means of administering insulin. More than half of the patients in the North American study had never experienced severe hypoglycemia, despite a median duration of diabetes of 17 years (IQR = 9 to 25 years). Only 7% of patients in this study experienced severe hypoglycemia in the year prior to the study. The proportion of patients with reduced hypoglycemia awareness was low (4% to 16%) across the adult studies.

The age of children in the pediatric study (IGBB) ranged from 4 to 17 years. Most children were between the ages of 6 and 13, with six children with ages between 4 to 5 years and eight children between the ages of 14 and 17. More than half of the children in this study used an insulin pump as the primary means of administering insulin. A majority of children in the study had never experienced severe hypoglycemia.

Adult Studies

Pediatric Study

Edinburgh Hypoglycemia Scale

Symptoms of hypoglycemia were assessed in the adult studies using the Edinburgh Hypoglycemia Scale. The version of the scale used in the IGBI and IGBJ studies consists of 13 symptoms categorized into three subscales: cognitive dysfunction (inability to concentrate, blurred vision, anxiety, confusion, difficulty speaking, and double vision); neuroglycopenia (drowsiness, tiredness, hunger, and weakness); and autonomic symptoms (sweating, trembling, and warmness).

Each of the 13 symptoms could be scored as follows:

• 1 = not experiencing this (no symptom as all)
• 2 = only experiencing a very mild case of this and it is easily tolerated
• 3 = only experiencing a mild case of this and it is tolerated
• 4 = experiencing a mild to moderate case of this and it is tolerated
• 5 = experiencing a moderate case of this and it is tolerated
• 6 = experiencing a moderate to severe level of this symptom; it is bothersome but tolerable
• 7 = experiencing a severe level of this symptom; it is hard to tolerate

The maximum total score is 91 and the maximum subscale scores are 42 for the cognitive dysfunction subscale, 28 for the neuroglycopenia subscale, and 21 for the autonomic symptoms subscale. The version of the scale used in Study IGBC was not described in detail in the clinical study report or in the corresponding publication.

The symptoms used in the 13-symptom version have been shown to be specific to hypoglycemia in an analysis of nine previous studies in a mixture of patients with T1D and normal hypoglycemia awareness (N = 92) and persons without diabetes (N = 77).¹⁸ Hypoglycemia was induced by insulin infusion or hyperinsulinemic glucose clamp and a previous version of the Edinburgh Hypoglycemia Scale¹⁹ was used to assess symptoms on similar rating scales.¹⁸ There were 13 symptoms common to all nine studies and their intensities and frequencies were similar regardless of diabetic status or hypoglycemia induction method.¹⁸ Principal components analysis was used to obtain the three subscales and these formed the basis of the 13-symptom scale.¹⁸ Most of the studies assessed cognitive function with mental performance tests, and symptoms in the cognitive dysfunction subscale may have been more prominent (compared with the non-specific neuroglycopenic symptoms) when patients were asked to perform mental tasks.¹⁸ A minimal important difference for the total score or subscale scores of any version of the Edinburgh Hypoglycemia Scale was not found.

Adverse Event Monitoring of Nasal and Non-nasal Symptoms

A scoring system that included nasal (rhinorrhea, nasal stuffiness/congestion, nasal itching, and sneezing) and non-nasal (itching/burning eyes, tearing/watering eyes, redness of eyes, and itching of ears or palate) symptoms were individually graded using a four-point scale approximately 15, 30, 60, and 90 minutes after each glucagon administration. The following scoring system was used during this study in order to quantify nasal symptoms:

• 0 = I am not experiencing this (no symptoms at all).
• 1 = I am only experiencing a mild case of this and it is easily tolerated.
• 2 = I am only experiencing a moderate level of this symptom. It is bothersome but tolerable.
• 3 = I am experiencing a severe level of this symptom. It is hard to tolerate and interferes with my activities.

Adult Studies

Primary outcome analysis approaches are described for the individual studies below and were similar across the studies. Sample size calculations for Study IGBC to test the noninferiority of intranasal glucagon treatment and intramuscular glucagon treatment among subjects with T1D used the following assumptions: 80% power; a response rate of 95% for both treatments; a noninferiority limit of 10 percentage points (absolute value); a one-sided alpha level of 0.025; and a correlation of zero. Given these assumptions, the sample size required was 75 participants with T1D. An additional seven participants with T2D were also to be enrolled but were only analyzed as exploratory analyses. Sample size calculations for Study IGBI for the same primary outcome included the following assumptions: 90% power; a treatment success rate of 98% for both treatments; a noninferiority margin of 10%; a two-sided alpha level of 0.05; and a within-patient correlation of zero between two treatment visits. Therefore, assuming a 5% dropout rate, the study planned to enroll 70 patients with a target of having at least 66 patients with evaluable data from both treatment visits. Sample size calculations for Study IGBJ for the same primary outcome included the following assumptions: 90% power; a treatment success rate of 98% for both treatments; a noninferiority margin of 10%; a one-sided alpha level of 0.025; and a within-patient correlation of zero between two treatment visits. Therefore, assuming an approximately 10% dropout rate, the study planned to enroll 75 patients with a target of having at least 66 patients with evaluable data from both treatment visits.

Noninferiority of nasal glucagon was declared when the upper limit of the two-sided 95% CI of the mean difference in proportion of patients with the primary outcome of success was less than the noninferiority margin of 10%. The sponsor stated that its selection of a noninferiority margin of 10% for all studies was based upon data from a simulated emergency study in which 10% of the patients (parents of children and adolescents with T1D) failed entirely to administer injectable glucagon.^7,20

The individual components of the primary outcome were summarized but no formal statistical testing was performed on these data. There was also a subgroup analysis of the time to primary outcome in patients with type 1 and type 2 diabetes, but these subgroups were very small and there was no formal testing done for statistical interaction by diabetes type.

The statistical analysis plans for Studies IGBC, IGBI, and IGBJ indicated that Kaplan–Meier curves would be constructed for the time-to-primary-outcome analyses, using standard censoring techniques. A treatment group comparison of the time from treatment to primary outcome was also completed, using the marginal Cox proportional hazard models for clustered data (to account for the correlation due to the crossover design), adjusted for central lab nadir blood glucose and treatment period. For Study IGBC, the P value of the treatment arm comparison of the time from treatment to outcome was derived using the marginal Cox proportional hazards model for clustered data. The log rank test was used to assess these data in Studies IGBI and IGBJ. There was no statistical testing for carryover effects. All analyses on the primary end point were conducted in the population of patients who completed both treatment visits and who also had evaluable data (i.e., the per-protocol population). Sensitivity analyses were conducted using the population of patients who were randomized (i.e., the intent-to-treat population).

A treatment comparison of the Edinburgh Hypoglycemia Scale score at each time point after glucagon administration was completed using linear mixed models with repeated measures adjusting for the treatment period and score at visit arrival.

Study IGBC: A one-sided 97.5% CI was obtained from the one-sample mean of the paired differences in the primary outcome of success. Noninferiority of intranasal glucagon was declared if the upper limit of the one-sided 97.5% CI constructed on the difference in proportions (intramuscular glucagon:intranasal glucagon), was less than the noninferiority limit of 10%. The difference in the proportion of successes between the treatment arms and the one-sided 97.5% CI was also calculated using a Poisson regression model, incorporating a generalized estimating equation with adjustments for nadir glucose and the treatment period (i.e., first treatment for participant versus second treatment for participant). The primary analysis for IGBC excluded six patients with T2D. For serial measurements of plasma glucose the analysis was completed using a linear mixed model with repeated measures that accounted for the correlation due to the crossover design and the correlation due to multiple measures, adjusting for starting glucose level and time period.

Summary statistics for plasma glucose concentrations at each time point across the dosing visit were calculated with imputation for missing glucose values and glucose values after receipt of intervention treatment using Rubin’s multiple imputation method, based on available glucose measurements and treatment arm. A treatment comparison of the blood glucose concentration over the 90 minutes after administration of glucagon was performed using a linear mixed model with repeated measures adjusted for nadir glucose and time period.

Studies IGBI and IGBJ: For the primary outcome, a two-sided 95% CI was obtained from the one-sample mean of the paired differences in the primary outcome (1 = outcome observed; 0 = outcome not observed) across the two treatment visits. For each patient the paired difference of treatment success between intramuscular glucagon and nasal glucagon was calculated and a t-test was used to create the 95% CI of the mean difference.

Descriptive statistics were used to summarize the serial measurements of plasma glucose. A between-treatment comparison of baseline and post-dose plasma glucose values over the 90 minutes of the post-dose period (240 minutes for Study IGBJ) was performed using a linear mixed model with repeated measures. This model accounted for the correlation due to the crossover design and the correlation due to multiple measures. It included baseline and treatment period time points and their interaction as covariates. Least squares means and two-sided 95% CIs were calculated for the difference in plasma glucose between treatment groups at each time point.

Pediatric Study

Analysis Populations

The main analyses for the primary outcome for the three adult studies was a per-protocol analysis based on the population of patients who completed two treatment visits and who had evaluable data from those visits. Evaluable data meant that these patients received glucagon and had no rescue treatment for severe hypoglycemia prior to or within the first 10 minutes after glucagon administration. The main analyses for Study IGBC excluded the six patients with T2D. In the adult studies, secondary outcome analyses were performed in the population that received at least one dose of study drug.

Safety analysis populations included all data from dosing visits where glucagon was received.

The pharmacokinetic and pharmacodynamic analyses in the pediatric study included patients who provided evaluable data for at least one treatment.

Rescue Oral Carbohydrates

▬▬▬▬▬

▬▬▬▬▬

▬▬▬▬▬

Treatment Success

The primary outcome of the three adult studies was treatment success defined as a plasma glucose increase to greater than or equal to 3.9 mmol/L or an increase of greater than or equal to 1.1 mmol/L from nadir 30 minutes after glucagon administration (Table 8). The rates of treatment success were 100% for both intranasal and intramuscular glucagon in both the IGBI and IGBJ studies. The treatment success rate was also 100% for intramuscular glucagon in the IGBC study but not for intranasal glucagon in Study IGBC, which had a success rate of 99%. The results of the primary outcome in all three adult studies met the pre-specified criteria for noninferiority since the upper boundary of the CIs did not exceed 10% in any of the three adult studies.

Table 8

Summary of Treatment Success.

In the pediatric trial IGBB, treatment success rates were 100% in all treatment groups.

The two components of the primary outcome were also assessed individually (plasma glucose increase to ≥ 3.9 mmol/L or plasma glucose increase of ≥ 1.1 mmol/L from nadir). For Study IGBC, the success rates of the individual components were similar between the intranasal and intramuscular treatments. The success rates for Studies IGBI and IGBJ were 100% for both intranasal and intramuscular glucagon.

Time-to-Treatment Success

Time-to-event analyses were performed on the primary outcome in all studies. The mean time-to-treatment success in the adult studies ranged from 10 to 16 minutes (Table 8, Figure 5, Figure 6, andFigure 7). The mean time-to-treatment success was longer with intranasal glucagon relative to intramuscular glucagon in Studies IGBC and IGBI. The difference between the treatments was approximately four minutes in Study IGBC and 1.6 minutes in Study IGBI (variance not reported) and the difference was statistically significant in both studies. ▬▬▬▬▬. There was no statistically significant difference in time-to-treatment success in Study IGBJ.

Figure 5

▬▬▬▬▬. ▬▬▬▬▬

In Study IGBB, post hoc analyses of time-to-treatment success were similar between intranasal and intramuscular glucagon groups.

Symptoms of Hypoglycemia

The Edinburgh Hypoglycemia Scale was used to assess severity of hypoglycemic symptoms at the time of administration of glucagon and for up to 60 minutes after. The maximum total score is 91 and the maximum subscale scores are: 42 for the cognitive dysfunction subscale; 28 for the neuroglycopenia subscale; and 21 for the autonomic symptoms subscale. A higher score indicated greater severity of symptoms. Arithmetic means were provided for Study IGBC and least squares means were provided for Studies IGBI and IGBJ (Table 9); the reason for this was not provided by the sponsor.

Table 9

Edinburgh Hypoglycemia Scale Total Score Summary.

In Study IGBC, the scores were higher (worse) at all time points after glucagon was administered for the intranasal glucagon treatment compared to the intramuscular glucagon treatment, and the differences were statistically significant at 15, 30, 45, and 60 minutes. ▬▬▬▬▬.

There were no data available for the protocol-specified efficacy outcomes of time to dose administration, quality of life, or measures of caregiver and patient satisfaction (as listed in Table 4).

Adverse Events

Across the adult studies the proportion of patients reporting at least one AE ranged from 19% to 57% after receiving either intranasal glucagon or intramuscular glucagon (Table 10). The overall rates of AEs were similar between the two treatments. The most frequently reported AEs included nausea, vomiting, headache, nasal discomfort, nasal congestion, increased lacrimation, fatigue, nasopharyngitis, and upper respiratory tract irritation. Oropharyngeal and eye symptoms occurred more frequently in patients after receiving intranasal glucagon compared to intramuscular glucagon in Study IGBC. This included nasal discomfort (10% with intranasal versus 0% with intramuscular), nasal congestion (8% intranasal versus intramuscular 1%), lacrimation increased (intranasal 8% versus intramuscular 1%), upper respiratory tract irritation (intranasal 19% versus intramuscular 1%). Nasal itching (49%) and sneezing (24%) occurred more frequently in Study IGBI after treatment with intranasal glucagon compared to intramuscular glucagon (0%).

Table 10

Summary of Harms in Adult Studies (Safety Population).

▬▬▬▬▬⁴ The proportion of patients reporting at least one AE in the pediatric study ranged from 42% to 100% across the different age subgroups after receiving either intranasal glucagon or intramuscular glucagon. The most commonly reported AEs after receiving intranasal glucagon were nausea (8% to 23%), vomiting (17% to 50%), and headache (8% to 33%).

Serious Adverse Events

▬▬▬▬▬

In the pediatric study (IGBB), one seven-year-old male child experienced a SAE of hypoglycemia during induction of hypoglycemia with insulin. The patient made a full recovery after receiving oral carbohydrates.

Withdrawals Due to Adverse Events

There were two withdrawals due to the AE of vomiting in the adult studies that occurred in relation to receiving intranasal glucagon.

Mortality

There were no deaths in the studies.

Table 11

Summary of Harms in Pediatric Studies.

Internal Validity

It was not possible to assess the balance of some baseline prognostic factors between the randomized groups since baseline characteristics were often reported for the entire patient population and not for the separate groups based upon sequence (e.g., intranasal followed by intramuscular versus intramuscular followed by intranasal).²¹ Where these were reported, there appeared to be adequate balance of prognostic factors between the randomized groups. For example, the glucose level at nadir (Table 6) was balanced between the groups.

Table 6

Summary of Baseline Characteristics.

The investigators suggest that delays in plasma glucose response to intranasal glucagon would be offset by reductions in administration time relative to intramuscular glucagon.²² Reviewers agree that this this is a rational hypothesis, but it is based on indirect evidence and has not been directly demonstrated in the intranasal glucagon trials performed to date.

The primary outcome for Study IGBB (children) does not appear to have been defined a priori. The publication states that the primary outcome was a greater than 1.4 mmol/L rise in plasma glucose within 20 minutes after glucagon administration, but this is not stated in either the sponsor’s statistical analysis plan or in the trial registry.^9–11 There was no formal sample size calculation performed for this study. For this reason, it is not known if intranasal glucagon is noninferior to intramuscular glucagon in the pediatric population. Although this study was designed to assess the pharmacokinetics and pharmacodynamics of intranasal glucagon relative to intramuscular glucagon, there were other inconsistencies about the way the study was reported in the publication compared to the clinical study report. For example, the publication referred to it as a phase I study and the clinical study report classified it as a phase III study. The subgroup analyses were predefined in the study protocol, but the number of children in the subgroups by age were small, and for these reasons the data from this trial cannot be considered conclusive evidence of intranasal glucagon efficacy and its harms relative to intramuscular glucagon.

▬▬▬▬▬

All studies were open label, and this could have impacted assessment of subjective outcomes such as the AEs and Edinburgh Hypoglycemia Symptoms.

External Validity

The primary limitation of the four trials that met the inclusion criteria for the systematic review is that the trials did not attempt to mimic real-world conditions. The study medications were administered under controlled conditions by trained health professionals, and not by caregivers or bystanders who may not have the same level of training in patient assessment and administration technique. Hypoglycemia was induced and symptom criteria for hypoglycemia were not used in the protocol to induce hypoglycemia. Achieving hypoglycemia was based on glucose levels alone. Intranasal glucagon is indicated for treatment of severe hypoglycemic reactions, but the controlled trials were not designed to study recovery from severe hypoglycemia. Reviewers acknowledge that real-world studies including the conditions specified in the indication (e.g., impaired consciousness) would be difficult to achieve; however, a major limitation of the studies remains since there were no controlled trials that tested the product under the conditions specified in the indication. Given the uniformity of the pharmacodynamic response to exogenous glucagon, reviewers believe that the extrapolation of the results of the trials to severe hypoglycemia is reasonable, but there remains uncertainty about the time to response relative to intramuscular glucagon since this has not been directly quantified under severe hypoglycemic conditions.

According to the clinical expert consulted for this review, the populations enrolled in the clinical trials are reasonably similar to the Canadian patients who would be prescribed glucagon nasal powder. Most patients were white, non-Hispanic, or Japanese, but there are no known reasons to believe that intranasal glucagon would have variable effects based on ethnicity. For example, the clinical expert confirmed that the rates of hypoglycemia awareness reported in the adult trials were similar to what would be expected in the Canadian population of adults with diabetes. There were some groups under-represented in the trials. Trials lacked older (e.g., > 65 years) populations who would be expected to have a longer duration of disease and a higher proportion of individuals with reduced awareness of hypoglycemia. Trials also enrolled lower numbers of patients with T2D, although it would be expected from the T2D clinical trial data and underlying physiology that the response to glucagon would be similar in this patient population.

The primary outcome of the three adult trials was resolution of low glucose levels within a 30-minute interval. Clinicians and patients would expect a resolution of low glucose levels in less than 30 minutes given the serious sequelae that can result from severe hypoglycemia that is not promptly and successfully treated.

Methods

The B001 and B002 studies were multi-centre, single-arm, open label studies. Study B001 was conducted in 2015 at three centres in the US, and Study B002 was conducted from 2014 to 2015 at three centres in the US and six centres in Canada. In both studies, one centre was excluded from efficacy analyses due to non-compliance with good clinical practice (GCP). Study B002 was paused due to an issue with powder aggregation and resulting underdosing in some patients. An evaluation period of approximately six months was expected for both studies to reach the required sample size of evaluable events. Patients continued in the study until one or more hypoglycemic events occurred or the study was complete, whichever occurred first. Patients (and caregivers in Study B001) attended study visits two and four months following enrolment, as well as at the end of the study.

Populations

Patients in both studies were required to have had T1D for more than one year and be in good general health to be included. Patients in Study B001 were at least four years of age and under 18 years of age and living with at least one caregiver. Patients in Study B002 were adults of 75 years of age or younger, had a body mass index between 18.5 and 35.0 kg/m², and were living with or in frequent contact with at least one caregiver. Patients in both studies were excluded if they had pheochromocytoma or insulinoma, or were using systemic beta blockers, indomethacin, warfarin, or anticholinergic drugs.

Patients in the efficacy analysis population (EAP) of Study B001 (see the statistical analysis section below for definitions) had a mean age of 10.2 years and a mean duration of diabetes of 6.3 years (Table 13). Most patients used an insulin pump as their primary insulin modality. In this population, 42.9% of patients had never experienced a severe hypoglycemic event and 21.4% had reduced hypoglycemia awareness according to the Clark Unawareness Score. For those patients who had experienced a severe hypoglycemic event in the past year, all patients had experienced it within the last 90 days.

Table 13

Summary of Baseline Characteristics (B001 and B002 Studies).

Patients in the Study B002 safety population (see the statistical analysis section below for definitions) had a mean age of 46.2 years and a mean duration of diabetes of 26.3 years (Table 13). Approximately half used an insulin pump and half used insulin injection as their primary insulin modality. In this population, 9.5% of patients had never experienced a severe hypoglycemic event and 40.6% had reduced hypoglycemia awareness. A severe hypoglycemic event had occurred in the past year in 58.2% of patients.

Interventions

In both studies, each patient was dispensed four doses of intranasal glucagon 3 mg and patients and caregivers were trained in its use. Patients and caregivers were also encouraged to keep one dose and one set of questionnaires with them at all times and the other doses and questionnaires in convenient locations. Patients in Study B001 were limited to four doses while patients in Study B002 could be dispensed additional doses.

Evaluable Events

Both moderate and severe hypoglycemic events were to be treated with intranasal glucagon and the definitions of each differed between the two studies. For hypoglycemic events to be considered evaluable, patients had to refrain from ingesting carbohydrates or injecting glucagon before responding or within 30 minutes of intranasal glucagon administration and not require external professional medical assistance. Events from centres with GCP non-compliance or occurring during the study pause in Study B002 were not considered evaluable events.

In Study B001, severe hypoglycemia was defined as the patient having severe neuroglycopenia (described as “usually resulting in coma or seizure”) requiring treatment with parenteral glucagon or IV glucose. In Study B002, severe hypoglycemia was defined as clinical incapacitation of the patient (i.e., unconscious, convulsing, or with severe mental disorientation) to the point where they required third-party assistance to treat the hypoglycemia.

Moderate hypoglycemic events in Study B001 were those in which the patient had signs and/or symptoms of neuroglycopenia and a blood glucose level of 70 mg/dL (equivalent to 3.9 mmol/L) or less at or near the time of treatment. The definition was similar in Study B002, except that the blood glucose level threshold was “approximately” 60 mg/dL (3.3 mmol/L) or less and did not appear to be strictly enforced.

Outcomes

The primary end point in both studies was originally the proportion of patients awaking or returning to a normal status within 30 minutes following study drug administration. The end point was amended to the proportion of hypoglycemic events rather than the proportion of patients.

Secondary end points in both studies included time to administer study drug, caregiver degree of satisfaction, and delivery method preference assessed with a hypoglycemia episode questionnaire completed by the caregiver. A set of pre-specified treatment-emergent AEs was assessed in the hypoglycemia episode questionnaire and could be recorded up to five hours post-administration in Study B002. More targeted AEs were assessed using a nasal score questionnaire, which was completed by caregivers in Study B001 and patients in Study B002. A tertiary end point of change in blood glucose level from time of study drug administration to 15, 30, and 45 minutes following administration was reported.

Statistical Analysis

No statistical tests were performed on the data. Efficacy end points were evaluated in the EAP, defined as enrolled patients who received at least one dose of study drug in an evaluable event and with evaluable information on treatment response. Questionnaire-based outcomes were reported for the EAP in Study B001 and for the main safety analysis population (MSAP) in Study B002, which was defined as enrolled patients who received at least one dose of study drug and experienced at least one hypoglycemic event (patients from GCP non-compliant sites and those underdosed during the pause in Study B002 were excluded). A sensitivity safety analysis population (SSAP) was also defined, which consisted of all enrolled patients who received at least one dose of study drug.

Study B001 was designed to include approximately 20 events of severe or moderate hypoglycemia. Study B002 targeted a sample size of 129 events of severe or moderate hypoglycemia, assuming that 75% of events would involve a successful response. The sample size was selected to yield a 95% CI for the primary end point with a width of 15%.

Patient Disposition

Details on patient disposition in the studies are provided in Table 14. Since patients could have experienced hypoglycemic events prior to discontinuation, early discontinuation did not necessarily exclude patients from efficacy analyses.

Table 14

Patient Disposition (B001 and B002 Studies).

Outside of the GCP non-compliant centre, there were no protocol deviations in Study B001 that were considered by the sponsor as likely to have affected the results or conclusions. In Study B002, seven severe hypoglycemic events were excluded from the EAP due to: dose administration before the study pause (n = 1); consumption of oral carbohydrates (n = 2); not fully depressing the device plunger (n = 2); and GCP non-compliance (n = 2). In the total pool of hypoglycemic events, 22 events were excluded from the EAP for the above reasons, including seven events during which the device plunger was not fully depressed. Other reasons for exclusion were primary outcome data missing; patient was found to be ineligible; and device was triggered in the air (n = 1).

Treatment Exposure

▬▬▬▬▬

Table 15

Treatment Exposure (B001 and B002 Studies — Safety Population).

Efficacy

Response To Study Drug Administration

Results for the primary end point and time to response in evaluable hypoglycemic events are presented in Table 16. In Study B001, there were 33 evaluable events (all of them moderate hypoglycemia) in 14 patients. Blood glucose recorded at the time of glucagon administration ranged from 2.3 to 3.9 mmol/L. In all events, the patient returned to normal status within 30 minutes of study drug administration.

Table 16

Summary of Hypoglycemic Events and Resolution (B001 and B002 Studies).

In Study B002, there were 157 evaluable events of moderate or severe hypoglycemia in 69 patients and the patient awoke (in cases of unconsciousness or convulsion) or returned to normal status in 96.2% of the events. In 3.4% of events, the patient returned to normal status after more than 30 minutes had elapsed following study drug administration, and in one event (0.7%) the patient did not return to normal status due to extreme headache. Blood glucose recorded at the time of glucagon administration ranged from 1.2 to 4.1 mmol/L. There were 12 events of severe hypoglycemia in seven patients (with one patient experiencing six of these events) and in all of these events the patients awoke or returned to normal status within 15 minutes of study drug administration, regardless of whether they were conscious at the time of administration.

Time To Administer Study Drug

According to the hypoglycemia episode questionnaire in Study B001, the time to administer the study drug (starting from when the device canister was opened) was less than two minutes in all evaluable events, with 60.6% of administration times being less than 30 seconds (Table 17). In Study B002, the time to administer the study drug was less than five minutes in all hypoglycemic events in the MSAP, with 70.4% of events having an administration time of less than 30 seconds.

Table 17

Selected Results From the User-Friendliness Questionnaire (B001 and B002 Studies).

Harms

The SSAP consisted of enrolled patients who received at least one dose of study drug, regardless of study site or, in Study B002, whether they were underdosed prior to the study pause. In Study B001, no patients in the SSAP (N = 22) reported a SAE and there were no deaths. In Study B002, one patient in the SSAP (N = 87) discontinued treatment due to an AE and there was one death from Klebsiella pneumoniae infection. Spontaneous AEs were not collected in either study.

Caregivers reported at least one AE in the hypoglycemia episode questionnaire for all patients in the EAP in Study B001 and 87.8% of patients in the MSAP in Study B002 (Table 18). The most commonly reported AEs were nasal discomfort/irritation (82.4% to 92.9%), watery eyes (85.7% in Study B001), and headache (54.1% to 71.4%). At least one AE that lasted for more than an hour was reported for half of the patients in Study B002.

Table 18

Summary of Questionnaire-Solicited Adverse Events (B001 and B002 Studies).

All caregivers for patients in the EAP in Study B001 and 72.4% of patients in the MSAP in Study B002 reported at least one AE in the nasal score questionnaire (Table 18). The most commonly reported AEs were runny nose (64.3% to 66.2%), watery eyes (55.4% to 78.6%), nasal congestion (36.5% to 50.0%), sneezing (33.8% to 50.0%), and nasal itching (28.6% to 56.8%).

Critical Appraisal

Overall, the B001 and B002 studies provide insight into the efficacy and harms of intranasal glucagon treatment in real-life events of moderate and severe hypoglycemia. The primary end point was assessed during real-life events as opposed to induced events of hypoglycemia. However, the results may overestimate the effectiveness of intranasal glucagon under real-world conditions due to the following factors: reporting of outcomes on a per-event basis rather than a per-patient basis, the more recent training of patients and caregivers in administering glucagon than would be expected in the real world, and the exclusion in Study B002 of events during which a full dose was not administered due to user error.

Methods

The IGBM and AMG111 studies were randomized, crossover, single-centre studies conducted in the US. Each study was conducted in two cohorts — caregiver-patient dyads, and acquaintance participants (APs) who were meant to represent non-caregiver bystanders. Details of the studies are provided in Table 12.

Caregiver-patient dyads attended three separate sessions spaced one week (Study AMG111) or at least one week (Study IGBM) apart. In the first session, patients received training on one glucagon device (randomized to injectable or intranasal) in the first session and subsequently trained their caregivers in the use of that device. In the second session, caregivers used the device in a simulated emergency situation, and patients received training in the other device and subsequently trained their caregivers in the use of the second device. In the third session, caregivers used the second device in a simulated emergency severe hypoglycemic event.

In these studies, APs attended either one session (Study AMG111) or two sessions spaced at least one week apart (Study IGBM). They were shown one glucagon device (randomized to injectable or intranasal) and then used it in a simulated emergency severe hypoglycemic event. This was repeated with the second device in the same session or in the second session.

Populations

Each caregiver participant (CP) in Study IGBM was a close friend, relative, or caregiver of the PWD and had not previously administered injectable glucagon or another rescue medication. In Study AMG111, CPs were the primary caregivers for the PWDs and had not previously used injectable glucagon and had not received recent training in the use of glucagon. The median age of CPs was 51 years (range of 18 to 75 years) in Study IGBM and 54 years (range of 20 to 69 years) in Study AMG111 (Table 19).

Table 19

Summary of Baseline Characteristics (Simulation Studies).

Most PWDs had T2D and the median duration of diabetes was 16 years in Study IGBM and 15 years in Study AMG111. In Study IGBM, PWDs were permitted to have been previously trained in the use of injectable glucagon, but not in the two years prior to the study. In Study AMG111, none of the PWDs had ever seen or received training on a glucagon device previously and none owned one at the time of the study.

APs were those who stated that they would try to help if an acquaintance experience a severe hypoglycemic event, and had no caregiving responsibilities to a PWD (Study IGBM) or had no experience with glucagon and diabetes (Study AMG111). The median age of APs was younger than those of CPs, being 41 years in Study IGBM and 40 years in Study AMG111.

Glucagon Administration Training

For each glucagon delivery device, PWDs in Study IGBM were trained for a maximum of 30 minutes by study personnel who reviewed the instructions for use and demonstrated the use of the device. PWDs were then given the device to verbalize and demonstrate understanding of the instructions. They also opened a new device and demonstrated the steps for administration, with study personnel correcting errors, reteaching missed steps, and answering questions on use of the device. After an hour-long break that included a distractor task for the PWD, the PWD relayed the device instructions to their caregiver and were allowed to show but not actuate a new glucagon delivery device (the intranasal device had to remain in its shrink-wrapped tube).

For each glucagon delivery device in Study AMG111, study personnel read the instructions for use to the PWDs and demonstrated the administration procedure without actuating the device. PWDs could handle the device but not actuate it. After a 10- to 30- minute break which included distractor tasks, PWDs then discussed how to use the device with their CPs. The intranasal device was not available for demonstration.

In both studies, APs received no training on the glucagon delivery devices. In Study IGBM they were given basic information about severe hypoglycemia, and in both studies APs were shown each device.

Simulated Severe Hypoglycemic Events

In both studies, CPs and APs participated in videotaped simulated severe hypoglycemic events in which they had to find a glucagon device (injectable or intranasal) and administer glucagon to a mannequin representing the person experiencing severe hypoglycemia. Device order for the sessions was randomized in Study IGBM, while device order in Study AMG111 appeared to be assigned according to whether the participant’s identification number in the study was odd or even.

In Study IGBM, the medical mannequin was clothed and had simulating breathing, blinking, pulse, heart sounds, and perspiration. Prior to starting the simulation, participants were informed that they would find a mannequin in the room that represented their associated PWD (for CPs) or a fictional co-worker (for APs). Participants were informed that the person had passed out due to hypoglycemia, and the importance of administering rescue medication was emphasized. They were also informed that their performance was being timed and recorded on video, that the glucagon rescue device would in the bedroom drawer (for CPs) or the mannequin’s backpack (for APs), and that the ambulance would not arrive for 15 minutes. Alongside the glucagon rescue device were other items, including diabetes supplies. For CP simulations, there was a television playing and a cell phone alarm sound. For AP simulations, there was a computer that was on and a cell phone alarm sound.

In Study AMG111, the mannequin was clothed. Prior to starting the simulation, participants were informed that the mannequin was in severe hypoglycemia and that they needed to find the glucagon rescue device in the mannequin’s backpack (which also contained diabetes supplies) and administer glucagon to the mannequin as quickly as possible.

In the first session, participants were told that they were being recorded on video and that the video was being streamed live over the internet and could appear in future educational and promotional material. They were also told that a team of experts was watching and evaluating them from behind a one-way mirror and the importance of administering rescue medication was emphasized. During the scenario, someone knocked loudly on the door and stated that they would make sure the ambulance was on its way once the participant found the glucagon device.

In the second session, participants were reminded of the situation and distractions were more frequent than in the first session. A loud beeping sound at one beep per second played throughout the simulation and increased in speed and intensity while study personnel made statements meant to simulate those of a distressed bystander if they deemed the participant was not engaging in the scenario. There were also distractions when the participants opened the glucagon packaging and at 30 seconds after the glucagon was found.

Outcomes

The primary end point in Study IGBM was the percentage of CPs who successfully administered a complete dose of glucagon, defined as at least 90% of glucagon drug solution for the injectable glucagon kit and the device plunger being fully depressed for the intranasal glucagon device, and completed all critical steps for the administration. The critical steps for the intranasal device were removing the device from packaging (which included shrink-wrap); not testing before use; inserting the device tip into one of the mannequin’s nostrils; and pushing the plunger (keeping the tip inside the nostril) until the green line no longer showed. The critical steps for the injectable glucagon kit were removing the device from packaging; injecting the diluent from the syringe into the vial containing drug powder; ensuring the drug powder was dissolved (by shaking and/or swirling); drawing the dissolved drug into the syringe; and injecting the drug into the mannequin at an appropriate site for intramuscular administration (thigh, buttock, or upper arm). The percentage of APs who successfully administered a complete dose of glucagon was a secondary end point. No primary end point was defined in Study AMG111, though the percentages of CPs and APs successfully administering a full dose of glucagon were reported, as well as the percentages of CPs and APs administering a partial dose of injectable glucagon. Partial dose administration was not possible with intranasal glucagon because the actuation mechanism ensured the entire dose was expelled.

Time to complete administration, starting from when the participant found the glucagon device and ending when the dose was administered, was measured in both studies. In Study IGBM, the simulation timer was stopped when the participant administered a dose of glucagon or after 15 minutes had elapsed.

Device preference was assessed using questionnaires in both studies for CPs, APs, and PWDs. Participants and PWDs in Study IGBM rated strength of preference on a 5-point Likert scale for the respective items. In Study AMG111, PWDs were asked to indicate which device they preferred and CPs and APs were asked to indicate the preferred device, with an option for no preference. Satisfaction was also assessed in Study IGBM.

Statistical Analysis

▬▬▬▬▬

Sample size considerations were not described for Study IGBM. In Study AMG111, sample sizes were based on the expected numbers of CPs and APs needed for 95% power to detect a within-subject difference of at least 40 seconds (with a standard deviation of 40 seconds) in time to administer study drug (intranasal versus injectable glucagon) at a significance level of 0.05. However, the sample size of 16 for each cohort was not reached due to many of the CPs and APs not completing administration of glucagon for both devices.

Patient and Participant Disposition

▬▬▬▬▬ In Study AMG111, 19 dyads and 20 APs were recruited. Of these, two dyads withdrew for personal reasons, one CP did not attempt the emergency simulations, and five APs either did not show up to sessions or withdrew from the study.

Table 20

Patient and Participant Disposition (Simulation Studies).

Efficacy

Drug Administration Success

Intranasal glucagon was consistently associated with higher rates of successful administration compared with injectable glucagon in CPs and APs in both studies (Table 21). In Study IGBM, a significantly greater percentage of CPs successfully administered intranasal glucagon versus injectable glucagon (90.3% versus 15.6%; P < 0.0001). In Study AMG111, a full dose was successfully administered by 94% of CPs for intranasal glucagon and 13% of CPs for injectable glucagon, while a partial dose of injectable glucagon was successfully administered by 38% of CPs.

Table 21

Drug Administration Success and Time to Administer Study Drug (Simulation Studies).

In Study IGBM, 90.9% of APs successfully administered intranasal glucagon and no APs successfully administered a full or partial dose of injectable glucagon (Table 21). In Study AMG111, 93% of APs successfully administered intranasal glucagon, no APs successfully administered a full dose of injectable glucagon, and 20% of acquaintance patients successfully administered a partial dose of injectable glucagon.

In Study AMG111, partial rather than full doses of injectable glucagon were administered by some participants due to failure to draw up all the solution into the syringe and/or failure to entirely depress the plunger.

Device Preference

In both studies, most CPs, PWDs, and APs expressed a preference for the intranasal glucagon device over the injectable glucagon kit (Table 22). In Study IGBM, 80.6% of CPs strongly preferred or preferred intranasal glucagon and 13.0% of CPs strongly preferred or preferred injectable glucagon. In terms of overall satisfaction, 74.2% of CPs strongly preferred or preferred intranasal glucagon and 9.7% preferred injectable glucagon. In Study AMG111, 87% of CPs preferred intranasal delivery of glucagon and 13% of CPs preferred needle-based delivery of glucagon for treating severe hypoglycemia.

Table 22

Selected Results From the Preference Questionnaire (Study IGBM).

In Study IGBM, 90.3% of PWDs strongly preferred or preferred intranasal glucagon in terms of feeling safe during a severe hypoglycemic event and 6.5% strongly preferred injectable glucagon. In Study AMG111, 69% of PWDs preferred intranasal delivery of glucagon and 19% preferred needle-based delivery of glucagon for the treatment of severe hypoglycemia by a third party.

In Study IGBM, 93.5% of APs strongly preferred or preferred intranasal glucagon and 3.2% strongly preferred injectable glucagon. In terms of overall satisfaction, 87.1% of APs strongly preferred or preferred intranasal glucagon and 3.2% strongly preferred injectable glucagon. In Study AMG111, all APs indicated that they would recommend that PWDs carry intranasal glucagon for the APs to treat them with (as opposed to injectable glucagon or neither device).

Critical Appraisal

While the differences in administration success and time to administration were pronounced and consistent between intranasal and intramuscular delivery in both studies, the generalizability of the results to the Canadian population of potential glucagon users is less clear.