Introduction

Today, the management of hyperglycemia is virtually possible for all patients with type 1 (T1DM) and type 2 (T2DM) diabetes mellitus, since treatment has substantially changed over the last 20 years for both kinds of patients. Oral agents (metformin) are used together with new oral (DPP-IV antagonists, SGLT-2 inhibitors) and injectable (GLP-1 agonists drugs), and insulin analogs are now in use. Prevention and treatment of hypoglycemia are still an unmet requirement, as euglycemia without hypo- or hyperglycemia is a major goal of management in patients with diabetes [1]. Hypoglycemia, symptomatic hypoglycemia, and severe hypoglycemia (a circumstance where the patient may be unconscious and requires the assistance of someone else), are frequent in T1DM and T2DM patients who use insulin or in T2DM patients on sulfonylureas; the frequency of hypoglycemia and severe hypoglycemia is greater in T1DM than in T2DM patients and depends on aggressive insulin treatment, on regimens of insulin administration, and on age [2,3,4,5,6,7,8,9,10,11]. Hypoglycemia is a risk factor for cardiovascular accidents [12], for falls and trauma [13], for cognitive impairment [14]; in addition, the experience of hypoglycemia leads to fear of hypoglycemia that in turn can limit optimal glycemic control in T1DM patients, children, adolescents, and adults [15, 16].

Treatment of hypoglycemia is still based on administration of carbohydrates or of glucagon [intramuscular (IM) or subcutaneous (SC) injection] [17]. The two approaches are similarly effective, as shown in a meta-analysis [18]. In 1983, it was shown that intranasal (IN) glucagon drops increased blood glucose levels in healthy volunteers [19]; later it was shown that IN glucagon solutions, IN glucagon sprays, IN glucagon powders are similarly effective, provided an enhancer is employed, as IN glucagon without enhancers is not absorbed through the nose and is without effects whatsoever [20,21,22,23,24]. During that decade, a few authors showed the efficacy of IN glucagon to resolve hypoglycemia in normal volunteers and in patients with diabetes, both adults and children [25,26,27,28]; a meta-analysis showed that IN glucagon and IM/SC glucagon were similarly effective in five studies performed with a total of 102 patients [18]. Based on evaluation of patient beliefs and patient expectations [28, 29], a Canadian pharmaceutical company in 2010 started to develop glucagon for IN administration. New studies have been published [30,31,32], leading to a total of 467 administrations in 269 T1DM patients. Food and Drug Administration (FDA) has approved the use of IN glucagon for treatment of severe hypoglycemia in T1DM patients on July 25, 2019 [33]. The aim of this systematic review and meta-analysis is to adjourn comparison of IN and IM/SC glucagon in resolution of hypoglycemia in insulin-treated T1DM patients [18] on a much greater number of patients.

Methods

In this meta-analysis we extended and deepened the methodology used in a previous meta-analysis [18]. Briefly, all studies comparing the hyperglycemic effects of IM/SC glucagon and of IN glucagon in T1DM patients during hypoglycemia were considered, published as full reports or in abstract form in any language up to April 2019. Retrieval of studies was based on MEDLINE, Cochrane Library, and EMBASE. A systematic literature search was conducted using the terms diabetes mellitus, hypoglycemia, insulin-induced hypoglycemia, and glucagon, limiting search to human clinical studies, including studies on children or elderly populations. Measure of efficacy was the number/percentage of individuals responding to IM/SC/IN glucagon as stated by the authors in each study.

Eight studies fulfilled the inclusion criteria [22, 24,25,26,27, 30,31,32]. All data are tabulated in Table 1. Appropriate methodology according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [34] was adhered to, as shown in flow diagram (Fig. 1). In all but one study, hypoglycemia was intentionally induced by insulin injection; in contrast, the study by Pontiroli et al. [24] evaluated inpatients randomly allocated to IN glucagon or IM glucagon under two pre-specified conditions: blood glucose levels < 70 mg/dL plus subjective feeling of hypoglycemia. Quality of reports and risk of bias were assessed according to RoBANS [35], that is, risk of bias linked to selection of participants, confounding variables, performance, detection and measurement of exposure, attrition, and reporting biases. A score was eventually built, indicating 0 = no risk, 1 = low risk, 2 = unclear risk, and 3 = high risk, based on the number of the above criteria available for each paper.

Table 1 Studies comparing effectiveness of intranasal (IN) and intramuscular/subcutaneous (IM/SC) glucagon (G) in resolution of hypoglycemia in subjects with insulin-induced hypoglycemia
Full size table
Fig. 1
figure 1

Flowchart of clinical trials included in the systematic review and meta-analysis. Studies dealing with glucagon-like peptides were excluded first; then, studies without original data (reviews, letters, and editorials) were excluded; finally, studies dealing with other actions of glucagon were excluded

Full size image

Statistical analysis

For each study, the number of individuals evaluated and the number of individuals with the increase in blood glucose (effective), with either glucagon IM/SC glucagon or IN glucagon, are reported, and the number of individuals not responding to either IN/IM/SC glucagon was calculated. Treatment outcomes were expressed as (ORs) with 95% confidence intervals (CIs) and pooled into an overall OR using a random-effects model according to DerSimonian and Laird [36]. The I2 index for heterogeneity was calculated for each outcome assessed, and potential sources of heterogeneity were discussed where appropriate. A p < 0.05 was considered indicative of statistically significant heterogeneity. Since all studies had administration of both IN glucagon and IM/SC glucagon, the number of individuals responding to glucagon was used to process the forest plot, using statistical program Stata 12 (Stata Corporation, College Station, Texas, USA) for MacIntosh. Because of the formula required to calculate ORs, when failures were null with both treatments, the studies were excluded from forest plots; therefore, a simulation was made, adding one failure for each arm in the remaining four studies (Fig. 2a and b, respectively).

Fig. 2
figure 2

Meta-analysis and publication bias of studies comparing intranasal and intramuscular glucagon/subcutaneous in the treatment of hypoglycemia. Since out of eight studies, there were only four reported failures, a simulation was made, adding one failure for each arm in the remaining four studies, and again there was no difference between IN and IM/SC glucagon. a Analyses derived from published data; b analyses from simulated data. Upper Panels forest plot of efficacy of intranasal glucagon compared to intramuscular glucagon in the treatment of hypoglycemia. Vertical line (1) represents no difference in the groups (OR); square and horizontal lines represent the point estimates and associated 95% CI for each comparison; the diamonds represent the pooled effect size, with the center representing the point estimate and the width representing the associated 95% CI Lower Panels funnel plot for detection of publication bias. The x-axis represents effect size as expressed by OR, and the y-axis represents standard error of OR. The vertical line represents the pooled effect size as calculated from the meta-analysis. The oblique dotted lines represent the 95% confidence limits around the summary treatment for each standard error on the vertical axis. These show the expected distribution of studies in the absence of heterogeneity or selection bias

Full size image

To explore the potential effect of patients or study characteristics on the pooled estimate of efficacy, we performed a meta-regression analysis. The dependent variable was failure in response from each study. The role of each covariate in heterogeneity was expressed by Wald test estimated by the meta-regression. The covariates considered for the meta-regression analysis were: number of participants of each study, age, and fasting glucose before IN and IM/SC glucagon. In a secondary analysis, we also evaluated the existence of a potential publication bias, defined as the tendency of authors and editors to handle studies in which the experimental results achieved statistical significance more favorably than in studies in which the results failed to reach significance, which would ultimately introduce bias into the overall published literature. Funnel plot asymmetry was evaluated by using the Egger’s test for small study effects through the meta-bias routine [37]. Discrepancies which arose during this process were resolved by discussion between the authors.

Results

Figure 2a-upper panel shows that the response to IN glucagon is not significantly different from the response to IM/SC glucagon. Failure to respond to IN glucagon was very rare, as it was to IM/SC glucagon. Since out of eight studies, there were only four reported failures [OR = 0.61 (95% CI 0.13–2.82)], a simulation was made, adding one failure for each arm in the remaining four studies, and again (Fig. 2b-upper panel) there was no difference between IN and IM/SC glucagon [OR = 0.80 (95% CI 0.28–2.32)]. Heterogeneity was low in the comparison and not statistically significant [I2 = 0.30 (df = 3) p = 0.960; I2 = 0.47 (df = 7) p = 1.000 in the simulation analysis], and no meta-regression was statistically significant (p always > 0.05, supplemental Figs. 1 and 2), meaning that the effect of IN and IM/SC glucagon was not dependent on size of the study, age of subjects, basal blood glucose levels prior to IN or IM/SC administration. The OR was 0.76 [95% CI 0.13–4.32] in young subjects and 0.80 [95% CI 0.18–3.45] in adults subjects (Supplemental Fig. 3). Publication bias was at all absent (Egger’s equation: p = 0.542; p = 0.365 in the simulation analysis). Quality of papers was generally good, and risk of bias was low (Table 2).

Table 2 Risk of bias of the studies included in this systematic review and meta < analysis
Full size table

Side effects of IN and injected (IM/SC) glucagon were similar, but were not analyzed systematically because of the different ways of reporting (numbers of subjects, grade of severity) in different studies; looking at single studies, one might conclude that the occurrence of gastrointestinal side effects (nausea and vomiting) and headache was somewhat lower with IN than with IM glucagon administration [27, 30,31,32], while local side effects (nasal irritation, ocular lacrimation) were few, but somewhat more with IN than with IM glucagon administration [25, 30].

Discussion

Administration of carbohydrates (oral or parenteral according to the level of consciousness) or glucagon [intramuscular (IM) or subcutaneous (SC) injection] is superimposable in efficacy in treatment of hypoglycemia [18]. In this paper, we have confirmed that IN glucagon is as effective as IM/SC glucagon in resolution of insulin-induced hypoglycemia in T1DM patients, as shown in a previous meta-analysis [18]. Interestingly, the meta-regression showed that the effect was not dependent on size of studies, age of patients, basal blood glucose prior to IN or IM/SC glucagon administration. This review indicates IN glucagon as a possible remedy for hypoglycemia in insulin-treated patients with type 1 diabetes. It seems wise to extrapolate that the rescue dose of IN glucagon might be 3 mg that is the dose approved in USA and Canada. Only one study explored different doses of IN glucagon, 2 and 3 mg, showing no differences in efficacy in children [31]. Absorption of IN-delivered glucagon requires an enhancer, and different promoters/enhancers were used in different studies, for instance glycocholic acid [19,20,21, 24, 25], didecanoyl-l-alpha-phosphatidyl-choline [26, 27], and beta-cyclodextrin plus dodecyl-phosphocholine [30,31,32]. Since IN glucagon was found to be as effective as SC/IM glucagon, it is unlikely that different enhancers had any impact on absorption and efficacy.

Besides efficacy, one should also consider the ease of IN administration as compared with the technical difficulties in preparing and administering IM/SC glucagon; as a matter of fact glucagon is not stable in solution and requires reconstitution before injection [38], a difficult maneuver for untrained patients and caregivers. This is also probably a reason why IM/SC glucagon is not frequently used, in spite of its efficacy in resolution of hypoglycemia [39, 40]. Both simulation studies and real-world experiences have shown that injected glucagon, in contrast to IN glucagon [41,42,43], is very difficult to administer, and a quite recent study showed that IN glucagon was similarly effective in the hands of caregivers of T1DM patients with or without training, while IM glucagon was much less effective in untrained than in trained caregivers [44]. Also of interest is the fact that the effect of IN glucagon is not altered by common cold or by the use of nasal decongestants [45].

This meta-analysis has limitations. The first limitation of this meta-analysis is that studies mostly came from a very few centers; the entity of this limitation is reduced by the overall results (lack of heterogeneity, no publication bias, quality of papers reporting studies). The second limitation is that, with the exception of study by Pontiroli et al. [24], studies were performed through insulin administration to induce hypoglycemia; this may not truly represent real-world use of the product, as patients in the trials received high doses of insulin to achieve hypoglycemia; this might have limited the ability to see a rise in glucose or determine effectiveness. The third limitation is that no study compared IN and IM/SC glucagon in unconscious patients; however, in the paper by Seaquist et al. [42] there were 12 episodes of severe hypoglycemia (unconsciousness, convulsions) that were rapidly and successfully treated by the patient’s caregivers using IN glucagon. Also, both in this paper and in the paper by Deeb et al. [43], people had the option to administer IM glucagon if they felt the patient was not responding to IN glucagon, and not a single participant had to receive IM glucagon.

Also, this meta-analysis was focused on studies performed in T1DM patients, and one might wander if differences in effect might exist between patients and healthy controls; as a matter of fact, Hvidberg et al. [46] compared IN and IM glucagon against insulin-induced hypoglycemia in healthy volunteers and found that the two ways of administration are equally effective in counteracting hypoglycemia, with no failures; this paper also showed that the glucose response was faster with IM than with IN glucagon, in agreement with the data of Rickels et al. [30] in adult patients with type 1 diabetes. Therefore, IN glucagon seems similarly effective in healthy volunteers and in adult T1DM patients. A preparation of IN glucagon has been approved by FDA [33] for the use by a third person (a caregiver); due to the ease of administration, and also of self-administration, some people might administer IN glucagon to themselves to treat an episode of mild hypoglycemia before it becomes an episode of severe hypoglycemia. This might contribute to reduce the fear of hypoglycemia, a condition has been associated with suboptimal diabetes management and health outcomes [47].

In conclusion, glucagon remains a key treatment for resolution of hypoglycemia in insulin-treated patients. IN glucagon is not inferior to IM/SC glucagon, at least in conscious patients, and has advantages represented by its ease of administration.