Early Loss of the Glucagon Response to Hypoglycemia in Adolescents With Type 1 Diabetes Open Access

Adolescents with type 1 diabetes were identified from the Western Australian Children’s Diabetes Database. Eligible patients aged 12–16 years attending the Diabetes Service at Princess Margaret Hospital for Children were approached to participate in the study. Exclusion criteria included a history of seizures not related to hypoglycemia, any episode of hypoglycemia (blood glucose level <3.5 mmol/L) in the 24 h preceding the study, and a history of severe hypoglycemia within the preceding 3 months. Clinical and demographic factors are prospectively recorded from diagnosis and recorded in the Western Australian Children’s Diabetes Database. These data were used to determine the factors included in the multivariate analysis. The control subjects were siblings of children with type 1 diabetes recruited from the community. The cross-sectional study design involved admission for a hypoglycemic hyperinsulinemic clamp study followed by an arginine stimulation test. The protocol was approved by the local ethics committee and signed informed consent was obtained for all subjects.

A modification of the glucose clamp technique described by Amiel et al. (8) was used. Patients were advised to fast for at least 8 h prior to admission to the research laboratory on the morning of the study. Only short-acting insulin was used in the 12 h preceding the study. Fasting samples were taken for plasma glucose, C-peptide, HbA_1c, and islet cell/glutamic acid decarboxylase antibodies.

The subjects were given an intravenous infusion of insulin (Humalog; Eli Lilly) at a continuous rate of 80 mU/m²/min, with doubling of the rate for the first 5 min. Plasma glucose was measured bedside at 5-min intervals and the glucose infusion rate (20% dextrose in saline) adjusted to achieve the target plasma glucose concentration. Blood samples were taken at 10- to 20-min intervals for measurement of counterregulatory hormones (glucagon, epinephrine, cortisol, and growth hormone) and free insulin levels.

The plasma glucose was initially stabilized between 5.0 and 5.5 mmol/L (90–108 mg/dL) for 60 min (euglycemic phase) and then reduced over a period of 30–40 min to achieve a target nadir of 2.8 mmol/L (50.4 mg/dL). Hypoglycemia was maintained for 40 min (hypoglycemic phase). On completion of the hypoglycemic phase, the insulin infusion was stopped and the plasma glucose was stabilized between 5.0 and 5.5 mmol/L (90–108 mg/dL) for at least 20 min.

To assess glucagon release to stimuli other than hypoglycemia and residual insulin reserve, arginine was used as an alternative stimulus (9,10). Baseline samples for plasma glucagon were taken at 5 (t = −5) and 2 min (t = −2) prior to infusion of 10% arginine solution (5 g in 50 mL 0.9% sodium chloride over 30 s) (11). Stimulated plasma glucagon was measured at 2, 3, 4, and 5 min postarginine (11).

Plasma glucose was assessed at bedside by the glucose oxidase method (YSI analyzer; YSI, Yellow Springs, OH). Blood for plasma glucagon was collected in EDTA with trasylol, immediately stored on ice, separated within 4 h, and stored at −70°C until assay. Under these conditions, glucagon immunoreactivity is preserved (12). Plasma glucagon was analyzed using a radioimmunoassay (Glucagon RIA; Linco Research, Inc., St. Charles, MO). The coefficient of variation for the glucagon assay was 13.9% at 51.5 pg/mL and 10.2% at 70.1 pg/mL.

Plasma epinephrine levels were measured by ELISA (Diagnostika GmbH, Hamburg, Germany), and samples were analyzed in duplicate according to the manufacturer’s instructions (13). Blood was collected into tubes containing EDTA preservative. Samples were stored in an ice bath until they were centrifuged to separate the plasma within 2 h and then stored at −70°C prior to assay. The interassay coefficient of variation (CV) at 10 pmol/L and 5,460 pmol/L were 2 and 5.5%, respectively.

We assessed residual β-cell function based on both fasting- and arginine-stimulated C-peptide values. Cortisol, growth hormone, and C-peptide were measured by immunoassays (Immulite; Diagnostic Products Corporation, Los Angeles, CA) (14).

Free and total insulin were measured by radioimmunoassay after precipitation of endogenous antibodies with polyethylene glycol. Analytical recovery of free insulin was 99.3%, of total insulin 96.4%. For free insulin, assay precision (CV) was 4.0–13.0% (intra-assay) and 7.8–10.7% (interassay); for total insulin, 3.6–9.5 and 6.6–11.7%, respectively (15). Glycated hemoglobin was measured by agglutination inhibition immunoassay (Ames DCA 2000, non–type 1 diabetes reference interval <6.2%).

Demographic data were expressed as mean and range. Duration of diabetes was also expressed as median and range to allow a more detailed analysis. All other results were described as mean and 95% CIs.

For the calculation of baseline and stimulated values, we used the average of time points 0–60 min for euglycemia and 130–150 min for hypoglycemia. A positive response to hypoglycemia was defined as a stimulated value greater than 3 SDs from the respective baseline during euglycemia. This approach was used by the Diabetes in Children Network Study Group (16) (see Table 2).

The glucagon responses to hypoglycemia and arginine were compared between the groups for each time point of analysis and are illustrated in Fig. 1A and B.

A linear mixed model was used for analysis of the glucagon responses. Stepwise linear regression was used in the subjects with type 1 diabetes to identify influencing factors. Statistical significance was set at P < 0.05 (Stata Statistical Software, release 12; StataCorp, College, Station, TX).

A total of 28 individuals with type 1 diabetes (14 females and 14 males) and 12 healthy volunteers (6 females and 6 males) were studied. Clinical characteristics are summarized in Table 1. All patients with type 1 diabetes were islet cell and/or glutamic acid decarboxylase antibody positive and on regular insulin therapy: short-acting insulin via insulin pump (n = 6) and combinations of long- and short-acting insulin using two (n = 7), three (n = 6), or four (n = 9) injections per day. One subject was taking thyroxine for autoimmune hypothyroidism. Median duration of type 1 diabetes was 0.66 years (range 0.01–9.9).

The glucagon peak to arginine stimulation was similar between groups (P = 0.27) (Fig. 1B). In contrast, the glucagon peak to hypoglycemia was reduced in the group with diabetes (95% CI): 68 (62–74) vs. 96 (87–115) pg/mL (P < 0.001) (Fig. 1A). A positive response was defined as greater than 3 SDs from baseline (25 pg/mL for control subjects and 48 pg/mL for the group with type 1 diabetes). Based on these values, only 7% of subjects with type 1 diabetes demonstrated a positive response in comparison with 83% of control subjects (Table 2). The response was lost at a median duration of diabetes of 8 months and as early as 1 month after diagnosis (R = −0.43, P < 0.01) (Table 3). Fasting and stimulated C-peptide were higher in the control group (Table 1).

Epinephrine, cortisol, and growth hormone responses to hypoglycemia were present, consistent with the generation of an adequate hypoglycemic stimulus (P < 0.05) (Table 2).

Stepwise linear regression revealed a significant correlation between glucagon response, duration of diabetes, and weight. There was no significant correlation with HbA_1c, BMI, fasting and stimulated C-peptide, height, and insulin level during hypoglycemia (Table 3).

This study demonstrates that in adolescents with type 1 diabetes, the glucagon response to an arginine stimulus is similar to healthy control subjects. However, adolescents with type 1 diabetes have a blunted glucagon response after hypoglycemia. Although this finding is not novel, this, to our knowledge, is the first study to show that the glucagon response to hypoglycemia was lost as early as 1 month after diagnosis and at a median duration of diabetes of 8 months. The glucagon response to hypoglycemia decreased with increasing duration of diabetes.

We assessed the counterregulatory response to hypoglycemia using clamp studies in adolescents with type 1 diabetes early in the course of the disease. Hoffman et al. (6) and Singer-Granick et al. (7) also demonstrated an impaired glucagon response to hypoglycemia in adolescents with type 1 diabetes. To further analyze and specify the participants’ response, Hoffman et al. (17) also included the pancreatic polypeptide release after hypoglycemia and the glucagon response to a mixed-meal stimulus. The interpretation of these studies, however, is limited by the methodology that used an insulin bolus to induce hypoglycemia. Compared with using clamp studies, this does not produce a reproducible hypoglycemic stimulus, resulted in a wide range of insulin levels, and is a technique in which the response to hypoglycemia cannot be compensated. This is why the stimulus may be modified. The Diabetes Research in Children Network Study Group demonstrated that the counterregulatory response can be lost in children aged 4–7 years with a diabetes duration of 3.3 ± 1.1 years (16). Hypoglycemia was induced using a subcutaneous insulin infusion protocol. In contrast to our study, that group focused on hypoglycemia-associated autonomic failure. Data on the glucagon release suggest a blunted response to hypoglycemia similar to our findings. However, in contrast to our study, this was not compared with a control group of healthy individuals (16).

As has been shown repeatedly in humans (5) and animal models (18), the preservation of the release of glucagon after arginine stimulation in all participants demonstrates that α-cells of patients with type 1 diabetes maintain their capability to release glucagon.

Importantly, the insulin infusion during clamp studies in itself can negatively impact the response to both hypoglycemia and arginine (19). Liu et al. (20) described that the glucagon response is suppressed by insulin levels of 125 mU/L. In a study by Diamond et al. (21), the glucagon response was negatively affected at insulin levels >279 mU/L. Mean insulin levels in this study were 91 and 83 mU/L in the control subjects and the type 1 patients with diabetes, respectively, and glucagon responses in the control group were greater after arginine than the hypoglycemic stimulus, suggesting that α-cell function was not significantly blunted by circulating insulin. Prior to the arginine stimulation, insulin was stopped and normoglycemia established.

The exact mechanisms regulating glucagon secretion in vivo remain to be identified. Several hypotheses have been proposed to explain the blunted response of glucagon release to hypoglycemia in type 1 diabetes patients. Changes to the intraislet insulin concentration as the result of β-cell loss is the most favored concept (22–24). Recent studies focused on the importance of zinc as a cofactor; however, this remains controversial (25,26). Looking at the central detection of hypoglycemia to induce and regulate counterregulatory responses, it appears from rat models that signal transduction in the ventromedial hypothalamus plays an important role. Proposed transmitters include glutamate, γ-aminobutyric acid, and catecholamines (27). They can be directly influenced by local glucose and insulin concentrations (28,29). In addition, the hypothalamus and brainstem might be involved as well (27). Other areas of research include the transcriptional control of pancreatic cell development and of the glucagon gene as well as proglucagon processing (30).

With respect to alternative stimuli of glucagon release, Pörksen et al. (31) conducted a large prospective study on the effect of a mixed-meal stimulus on glucagon release in adolescents with type 1 diabetes 1, 6, and 12 months after diagnosis. No association with residual β-cell function, age, and sex was noted, but blood glucose and glucagon-like peptide-1 were observed to be influencing factors. Similarly, Brown et al. (32) analyzed the change in glucagon response to a mixed-meal stimulus in young patients on five occasions during the first year after the diagnosis of diabetes. They observed a decline in C-peptide secretion paralleled by an increase in glucagon response over time, also suggesting that the α-cell secretory reserve had not been affected by the autoimmune process in type 1 diabetes. Neither study included an analysis of the response to hypoglycemia.

Several demographic and clinical factors were identified as predictors of the counterregulatory response to hypoglycemia. Consistent with our findings, the impact of duration of diabetes was reported previously. Most studies focused on adults (5,33,34). Results for children and adolescents vary; a loss of the counterregulatory response was detected as early as at the diagnosis of type 1 diabetes and shortly thereafter (6,35). Singer-Granick et al. (7) demonstrated a positive correlation with age but did not find a correlation between glucagon response and duration of type 1 diabetes in children and adolescents. This may be explained by a significant number of prepubertal participants, indicating a possible effect of puberty that could not be replicated in our group. Unlike Ross et al. (36) and Singer-Granick et al. (7), we did not find a relationship between glycemic control and the preservation of counterregulation in children and adolescents.

We conclude that adolescents with type 1 diabetes have impaired glucagon responses to hypoglycemia within 12 months of diagnosis. Prospective studies in children and adolescents starting from the onset of type 1 diabetes are needed to further characterize the mechanisms influencing the change over time in the glucagon response to hypoglycemia.

This study was funded by the Juvenile Diabetes Research Foundation.

No potential conflicts of interest relevant to this article were reported.

A.S., T.W.J., and E.A.D. developed the study protocol, researched data, contributed to discussion, and wrote, reviewed, and edited the manuscript. R.J.J. researched data, contributed to discussion, and reviewed and edited the manuscript. M.K.B. contributed to discussion and reviewed and edited the manuscript. P.O. researched data and reviewed and edited the manuscript. E.A.D. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank Leanne Youngs (Princess Margaret Hospital for Children) for her help with the clamp studies and work on the assays, and the diabetes clinical team at Princess Margaret Hospital for Children for their help with the recruitment of subjects. The authors also acknowledge the subjects and their families and thank them for their time and enthusiasm for this study.