Loss of the Decrement in Intraislet Insulin Plausibly Explains Loss of the Glucagon Response to Hypoglycemia in Insulin-Deficient Diabetes : Documentation of the Intraislet Insulin Hypothesis in Humans

For this study, 14 healthy young adults (7 women and 7 men) gave their written informed consent to participate. The study was approved by the Washington University Medical Center Human Studies Committee and conducted at the Washington University General Clinical Research Center. Subjects’ mean age (±SD) was 26 ± 4 years and their mean BMI was 23 ± 4 kg/m². All subjects had normal fasting plasma glucose concentrations, serum creatinine concentrations, and hematocrits and normal electrocardiograms (ECG).

Subjects reported to the research center early in the morning after an overnight fast on two occasions separated by at least 2 weeks. After normal blood pressures and heart rates in the supine and standing positions were documented, subjects assumed the supine position and remained in that position until the study was completed. Lines were inserted into an antecubital vein (for insulin and glucose infusions) and a dorsal hand vein, with that hand being kept in an ∼55°C Plexiglas box (for arterialized venous sampling) at t = ∼ −30 min. A hyperinsulinemic (2.0 mU · kg⁻¹ · min⁻¹)-euglycemic (∼5.0 mmol/l[90 mg/dl], 0–120 min) clamp followed by a hypoglycemic (∼3.0 mmol/l [55 mg/dl], 120–240 min) clamp was performed on both occasions. In random sequence, placebo or diazoxide (Proglycem; Schering-Plough, Kenilworth, NJ), 6.0 mg/kg, p.o. was administered at t = 0 min. Samples were obtained and blood pressures and heart rates (Propaq Encore; Protocol Systems, Beverton, OR) were recorded at t = −15, 0, 15, 30, 45, 60, 75, 90, 105, 120, 135, 150, 165, 180, 195, 210, 225, and 240 min. The ECG was monitored throughout. Neurogenic (autonomic) and neuroglycopenic symptoms of hypoglycemia were assessed at t = −15, 0, 30, 60, 90, 120, 150, 180, 210, and 240 min. Symptoms were quantitated by asking the subjects to score (from 0 [none] to 6 [severe]) 12 symptoms chosen based on our published data (14): 6 neurogenic (adrenergic: pounding heart, feeling shaky/tremulous, or feeling nervous/anxious; and cholinergic: feeling sweaty, hungry, or tingling) and 6 neuroglycopenic (difficulty thinking/feeling confused, and feeling tired, drowsy, weak, warm, faint, or dizzy) symptoms were assessed. Arginine hydrochloride (5.0 g) was injected intravenously after the 240-min sample; additional samples were obtained at t = 243, 245, and 247 min.

Plasma glucose concentrations were measured by the glucose oxidase method (Yellow Springs Analyzer 2; Yellow Springs Instruments, Yellow Springs, OH). Plasma insulin (33), C-peptide (33), glucagon (34), pancreatic polypeptide (35), growth hormone (36), and cortisol (37) levels were measured with radioimmunoassays. Plasma epinephrine and norepinephrine levels were measured with a single isotope derivative (radioenzymatic) method (38). Serum nonesterified fatty acids (39) and blood lactate (40) were measured with enzymatic techniques.

Data are given as mean ± SE, except where the SD is specified. Baseline adjusted data were analyzed by mixed-procedure, repeated-measures ANOVA. P values <0.0500 were considered to indicate statistically significant differences. Condition (diazoxide or placebo) or condition × time interaction ANOVA P values are reported. Increments in plasma glucagon under the two conditions were contrasted with a t test for paired data. Pearson correlation coefficients were determined for the relation between decrements in plasma C-peptide and increments in plasma glucagon.

Plasma insulin concentrations were similar and target plasma glucose concentrations were achieved during both clamp studies (Fig. 1). Plasma C-peptide concentrations were lower during the euglycemic phase of the diazoxide compared with the placebo clamp (P = 0.0015) (Fig. 2). Plasma C-peptide concentrations declined from 0.6 ± 0.1 nmol/l (1.7 ± 0.2 ng/ml) at t = 0 min to 0.4 ± 0.0 nmol/l (1.2 ± 0.1 ng/ml) at t = 120 min after placebo and from 0.6 ± 0.1 nmol/l (1.7 ± 0.2 ng/ml) at t = 0 min to 0.3 ± 0.0 nmol/l (0.8 ± 0.1 ng/ml) at t = 120 min after diazoxide administration. During the hypoglycemic phase, plasma C-peptide concentrations declined further in both clamps, to 0.1 ± 0.0 nmol/l (0.4 ± 0.1 ng/ml) with placebo and 0.1 ± 0.0 nmol/l (0.4 ± 0.1 ng/ml) with diazoxide (Fig. 2). Thus, during the induction of hypoglycemia, the decrement in mean C-peptide in the diazoxide clamps (0.4 ng/ml) was 50% of that in the placebo clamps (0.8 ng/ml) (P = 0.0015). It was notable that intravenous arginine did not raise C-peptide levels during hypoglycemia in either study (Fig. 2).

Plasma glucagon concentrations declined similarly during the euglycemic phase of both clamp studies (Fig. 3). Plasma glucagon concentrations declined from 34 ± 3 pmol/l (117 ± 9 pg/ml) at t = 0 min to 28 ± 2 pmol/l (98 ± 7 pg/ml) at t = 120 min with placebo and from 36 ± 3 pmol/l (124 ± 9 pg/ml) at t = 0 min to 29 ± 3 pmol/l (102 ± 9 pg/ml) at t = 120 min with diazoxide. The glucagon response to hypoglycemia was reduced (P = 0.0010) with the diazoxide compared with the placebo clamps (Fig. 3). Plasma glucagon concentrations rose to 43 ± 5 pmol/l (151 ± 16 pg/ml) at t = 240 min in the placebo clamps and to 35 ± 2 pmol/l (123 ± 8 pg/ml) at t = 240 min in the diazoxide clamps. Thus, during hypoglycemia, the increment in plasma glucagon in the diazoxide clamps (21 ± 5 pg/ml) was 45% of that in the placebo clamps (47 ± 11 pg/ml) (P = 0.027). It was notable that the glucagon response to intravenous arginine was not reduced by diazoxide (Fig. 3). The increments in plasma glucagon were related to the decrements in plasma C-peptide (r = −0.593, P = 0.015) during hypoglycemia; the correlation coefficient was −0.965 (P < 0.001) in the placebo clamps.

Plasma epinephrine (P = 0.1583) and norepinephrine (P = 0.2599) concentrations (Fig. 4), neurogenic symptom scores (P = 0.6859), plasma pancreatic polypeptide concentrations (P = 0.1456) (Fig. 5), and growth hormone and cortisol responses (Table 1) to hypoglycemia were not reduced in the diazoxide clamps. Indeed, the growth hormone response was enhanced (P = 0.0134). Cortisol levels were slightly lower during the euglycemic phase of the diazoxide clamps, but rose to levels comparable with those in the placebo clamps during hypoglycemia (P = 0.0255). Neuroglycopenic symptom scores were not altered (data not shown).

Heart rates and blood pressures were similar in both clamps (Table 2), although the mean heart rate tended to be higher (P = 0.1835) and the mean diastolic blood pressure tended to be lower (P = 0.0545) during the hypoglycemic phase of the diazoxide clamps. Blood lactate and serum nonesterified fatty acid levels (P = 0.6683) were similar under both conditions, although the increment in blood lactate during hypoglycemia was reduced slightly (P = 0.0154) in the diazoxide clamps (Table 3).

These data demonstrate, in healthy humans, that an ∼50% reduction in the decrement in insulin secretion, and thus in the decrement in intraislet insulin, during induction of hypoglycemia causes a >50% reduction in the glucagon response to hypoglycemia. Thus, they document the intraislet insulin hypothesis (18–23,30,31) for the signaling of the glucagon response to hypoglycemia in humans: a decrease in arterial glucose → a decrease in β-cell insulin secretion → a decrease in intraislet insulin → a decrease in tonic α-cell inhibition by insulin → an increase in glucagon secretion. Compared with placebo, diazoxide selectively, but only partially, suppressed plasma C-peptide concentrations, an index of insulin secretion, during the hyperinsulinemic-euglycemic phase of the clamps. C-peptide concentrations then decreased to comparable levels during hypoglycemia. Thus, there was a decrement in intraislet insulin during the induction of hypoglycemia and a subsequent increment in glucagon secretion during hypoglycemia in both the diazoxide and the placebo clamp studies. However, both the decrement in intraislet insulin during the induction of hypoglycemia and the subsequent increment in glucagon secretion during hypoglycemia were smaller in the diazoxide compared with the placebo clamps. Stated differently, a reduced signal (decrement in intraislet insulin) during induction of hypoglycemia led to a reduced response (increment in plasma glucagon) during hypoglycemia. The timing of the glucagon response was similar under both conditions, but the magnitude of the increment in plasma glucagon was related to the magnitude of the decrement in intraislet insulin.

The effect of diazoxide cannot be attributed to a direct inhibitory action of the drug on α-cell glucagon secretion. Baseline plasma glucagon concentrations and the brisk glucagon response to intravenous arginine were not reduced in the diazoxide, compared with the placebo, clamps. Furthermore, neither intravenous nor oral diazoxide suppressed plasma glucagon concentrations in healthy humans under nonclamped conditions (32). In addition, although by suppressing baseline insulin release diazoxide reduced the glucagon release response to low glucose from the perfused rat pancreas, consistent with the intraislet insulin hypothesis, it did not suppress basal glucagon release and enhanced, rather than inhibited, the glucagon release response to perfusion with high glucose (20). Finally, diazoxide did not decrease basal glucagon release from α-TC glucagonoma cells (41).

An important finding of our study, in the context of other signals for the glucagon response to hypoglycemia, was that the reduction of the intraislet insulin signal in the diazoxide clamps reduced the glucagon response to hypoglycemia, despite a low α-cell glucose concentration and an intact autonomic nervous system response. Activation of all three components of the autonomic nervous system—the adrenomedullary (plasma epinephrine and norepinephrine) (42), sympathetic neural (neurogenic symptoms) (42), and parasympathetic neural (plasma pancreatic polypeptide)—during hypoglycemia was similar in the diazoxide and placebo clamps. These data do not negate a role for autonomic activation in stimulating glucagon secretion during hypoglycemia (10–16). However, they do demonstrate a role for an intraislet insulin signal independent of the autonomic responses. The growth hormone and cortisol responses to hypoglycemia were also not reduced after diazoxide administration.

Several findings of this study, although not novel (2), are of interest. First, intravenous arginine did not stimulate insulin secretion during hypoglycemia under either clamp condition. This further illustrates that hypoglycemia suppresses insulin secretion, and does so potently. Second, the increase in plasma glucagon stimulated by intravenous arginine was followed by prompt increments in plasma glucose. This further illustrates that glucagon stimulates glucose production potently, despite substantial hyperinsulinemia. Third, as plasma glucose levels approached the physiological postabsorptive range, plasma epinephrine and norepinephrine levels fell. This further illustrates that the glycemic thresholds for catecholamine release, like those for other glucose counterregulatory hormones, lie below the physiological range.

Bingham et al. (43) found no significant effect of diazoxide on the glucagon response to hypoglycemia in healthy men, although the mean peak glucagon concentration tended to be lower after diazoxide than after placebo administration. However, they used a lower dosage of diazoxide and did not document an effect of the drug on insulin secretion by measuring plasma C-peptide concentrations. Furthermore, the precise temporal relation between diazoxide administration and induction of hypoglycemia is unclear in their report, and the frankly hypoglycemic glucose level (43 mg/dl) was maintained for only 40 min. In the present study, the effect of a reduced intraislet insulin signal to reduce the glucagon response became apparent during the 2nd h of hypoglycemia.

McCrimmon et al. (44) have reported that diazoxide injected bilaterally into the rat ventromedial hypothalamus increases the plasma epinephrine and glucagon responses to hypoglycemia. In the present study, systemic diazoxide administration did not alter the plasma epinephrine response and reduced the glucagon response.

Although the present data indicate that a decrease in intraislet insulin is a signal for glucagon secretion in response to hypoglycemia, a decrement in insulin secretion alone does not stimulate glucagon secretion. Plasma C-peptide concentrations declined during the hyperinsulinemic-euglycemic clamp, but plasma glucagon concentrations did not increase. In streptozotocin-induced diabetic rats, the switch off of superior pancreaticoduodenal artery insulin infusion elicited a glucagon response only when that was done at the time hypoglycemia was induced (22). Thus, there appears to be an interaction between a decrease in intraislet insulin and a decrease in α-cell glucose that triggers increased glucagon secretion (22,30).

In conclusion, the present data indicate that a decrease in intraislet insulin is a signal for the normal glucagon secretory response to hypoglycemia in healthy humans. The absence of that signal plausibly explains the loss of the glucagon response to falling plasma glucose concentrations, a key feature of the pathogenesis of iatrogenic hypoglycemia, in insulin-deficient (type 1 and advanced type 2) diabetes (3,4–9).

This study was supported, in part, by National Institutes of Health Grants R37-DK-27085, MO1-RR-00036, and P60-DK-20579, and a fellowship award from the American Diabetes Association.

The authors gratefully acknowledge the skilled assistance of the nursing staff of the Washington University General Clinical Research Center in the performance of these studies; the technical assistance of Krishan Jethi, Cornell Blake, Joy Brothers, Zena Lubovich, and Michael Morris; and the assistance of Janet Dedeke in the preparation of this manuscript.