Adrenergic Mediation of Hypoglycemia-Associated Autonomic Failure

This study was approved by the Washington University Human Research Protection Office and was conducted at the Washington University Clinical Research Unit (CRU).

Study participants comprised 17 adults (7 women, 10 men), with a mean (± SD) age of 29 ± 5 years and a BMI of 26.5 ± 4.6 kg/m², who gave their written consent. They were in good health as determined by medical history, physical examination, and fasting plasma glucose and creatinine concentrations, hematocrits, and electrocardiograms that were within normal reference ranges.

Participants were studied on 2 consecutive days on three occasions, separated by at least 2 weeks, after overnight fasts. Intravenous catheters were inserted into a hand vein, with that hand kept in a ∼55°C plexiglas box for arterialized venous blood sampling, and into a contralateral antecubital vein for insulin, glucose, and drug infusions and injections.

Day 1 involved hyperinsulinemic (2.0 mU/kg/min), euglycemic (90 mg/dL [5.0 mmol/L] × 1 h), and then hypoglycemic (54 mg/dL [3.0 mmol/L] × 2 h) clamps in the morning and again in the afternoon on all three occasions. Intravenous infusions during these day 1 glucose clamps were 1) saline, 2) the nonselective α-adrenergic receptor antagonist phentolamine mesylate (70 μg/kg, followed by 7.0 μg/kg/min) and the nonselective β-adrenergic antagonist propranolol hydrochloride (14 μg/kg, followed by 1.4 μg/kg/min), or 3) phentolamine plus propranolol and the muscarinic cholinergic antagonist atropine (1.0 mg injected intravenously at the end of each euglycemic clamp and 60 min into each hypoglycemic clamp) in random sequence. This experimental design is illustrated, with the actual plasma glucose concentration data, in Fig. 1.

Plasma glucose concentrations were measured every 5 min at the bedside (YSI Glucose Analyzer, Yellow Springs Instruments, Yellow Springs, OH) to guide 20% glucose infusions. Heart rates and blood pressures (GE Dash 3000, Fairfield, CT) were recorded at 15-min intervals, and the electrocardiogram was monitored throughout. Arterialized venous samples for the additional analytes detailed below were drawn at 15-min intervals during the morning euglycemic and hypoglycemic clamps on both days. This model of morning and afternoon hypoglycemia reduces the sympathoadrenal response to hypoglycemia the following day (1–4), and these doses of phentolamine, propranolol, and atropine are safe and hemodynamically, metabolically, and symptomatically effective (15).

Day 2 involved identical hyperinsulinemic euglycemic and then hypoglycemic clamps, with saline infusion, in the morning on all three occasions. Arterialized venous samples were again used to measure plasma glucose concentrations every 5 min and were drawn for the analytes detailed subsequently every 15 min.

Plasma glucose concentrations were measured with a glucose oxidase method (YSI Glucose Analyzer). Plasma insulin, C-peptide, growth hormone, and cortisol concentrations were measured with two site chemiluminescent assays (Immulite 1000, Siemens Corp., Los Angeles, CA). Plasma glucagon and pancreatic polypeptide concentrations were measured with Linco radioimmunoassays (Millipore, Temecula, CA). Plasma epinephrine and norepinephrine concentrations were measured with a single isotope-derivative (radioenzymatic) method (16).

Time and condition related variables were analyzed by repeated measures mixed model ANOVA. Values of P < 0.05 were considered to indicate statistically significant differences. Data are expressed as the mean ± SE, except where the SD is indicated.

Plasma glucose concentrations were clamped at euglycemic (∼90 mg/dL [5.0 mmol/L]) and then hypoglycemic (∼54 mg/dL [3.0 mmol/L]) levels in the morning and again in the afternoon—with intravenous infusions of saline, phentolamine plus propranolol (adrenergic blockade), or phentolamine plus propranolol with injection of atropine (adrenergic plus cholinergic blockade)—on day 1 and on the morning of day 2—with intravenous infusion of saline—after saline, phentolamine plus propranolol, or phentolamine plus propranolol with atropine on day 1 on three separate occasions (Fig. 1). The glucose infusion rates required to maintain the clamps were similar on all occasions (data not shown).

Compared with those during saline, increments in plasma epinephrine and norepinephrine concentrations during hypoglycemia were increased about threefold during adrenergic blockade and during adrenergic plus cholinergic blockade (both P < 0.0001; Fig. 2). The final epinephrine values were 352 ± 46 pg/mL (1,920 ± 251 pmol/L), 1,040 ± 160 pg/mL (5,680 ± 873 pmol/L), and 1,130 ± 193 pg/mL (6,170 ± 1,050 pmol/L), respectively. The final norepinephrine values were 257 ± 19 pg/mL (1.52 ± 0.11 nmol/L), 809 ± 84 pg/mL (4.78 ± 0.50 nmol/L), and 786 ± 100 pg/mL (4.65 ± 0.59 nmol/L), respectively. Increments in plasma pancreatic polypeptide concentrations during hypoglycemia were similar during saline and during adrenergic blockade but were prevented during adrenergic plus cholinergic blockade. The final pancreatic polypeptide values were 356 ± 45 pg/mL (85 ± 11 pmol/L), 374 ± 42 pg/mL (89 ± 10 pmol/L), and 75 ± 9 pg/mL (18 ± 2 pmol/L), respectively.

Plasma concentrations of infused insulin stabilized at ∼95 μU/mL (570 pmol/L) during infusion of saline but rose to ∼120 μU/mL (720 pmol/L) during adrenergic blockade and during adrenergic plus cholinergic blockade (both P < 0.0001; Fig. 3). The final insulin values were 96 ± 8 μU/mL (576 ± 48 pmol/L), 122 ± 12 μU/mL (732 ± 72 pmol/L), and 123 ± 9 μU/mL (738 ± 54 pmol/L), respectively. Increments in plasma glucagon concentrations during hypoglycemia were similar on all three occasions (Fig. 3). The final glucagon values were 98 ± 11 pg/mL (28 ± 3 pmol/L), 110 ± 10 pg/mL (32 ± 3 pmol/L), and 104 ± 7 pg/mL (30 ± 2 pmol/L), respectively.

Compared with those during saline, increments in plasma growth hormone and cortisol concentrations during hypoglycemia were increased during adrenergic blockade and adrenergic plus cholinergic blockade (P < 0.0001 under both conditions for both hormones; Fig. 3). The final growth hormone values were 13.3 ± 2.2 ng/mL (587 ± 97 pmol/L), 25.2 ± 3.2 ng/mL (1,110 ± 141 pmol/L), and 20.7 ± 2.8 ng/mL (914 ± 124 pmol/L), respectively. The final cortisol values were 18.4 ± 1.5 μg/dL (508 ± 41 nmol/L), 22.1 ± 1.4 μg/dL (610 ± 39 nmol/L), and 23.7 ± 1.6 μg/dL (654 ± 44 nmol/L), respectively.

Heart rate responses to hypoglycemia were decreased during adrenergic blockade and increased during adrenergic plus cholinergic blockade compared with those during saline (both P < 0.0001; Table 1). Systolic and diastolic blood pressures were reduced during adrenergic blockade and during adrenergic plus cholinergic blockade (both P < 0.0001; Table 1).

Compared with that during saline on day 1, the increment in the plasma epinephrine concentration during hypoglycemia was attenuated by ∼50% on day 2 (P < 0.0001; Fig. 4). The epinephrine values during the last hour of hypoglycemia were 351 ± 13 pg/mL (1920 ± 71 pmol/L) and 214 ± 22 pg/mL (1170 ± 120 pmol/L), respectively. In contrast, the increments in plasma epinephrine during hypoglycemia were not reduced on day 2 after adrenergic blockade or adrenergic plus cholinergic blockade on day 1 (Fig. 4). The epinephrine values during the last hour of hypoglycemia were 359 ± 26 pg/mL (1,960 ± 142 pmol/L) and 375 ± 11 pg/mL (2,050 ± 60 pmol/L), respectively.

Similarly, the increment in the plasma norepinephrine concentration during hypoglycemia was attenuated on day 2 compared with that during saline on day 1 (P < 0.0001; Fig. 4). The norepinephrine values during the last hour of hypoglycemia were 252 ± 4 pg/mL (1.49 ± 0.02 nmol/L) and 226 ± 7 pg/mL (1.34 ± 0.04 nmol/L), respectively. The increments in plasma norepinephrine during hypoglycemia were not reduced on day 2 after adrenergic blockade or adrenergic plus cholinergic blockade on day 1 (Fig. 4). The norepinephrine values during the last hour of hypoglycemia were 284 ± 11 pg/mL (1.68 ± 0.06 nmol/L) and 256 ± 11 pg/mL (1.51 ± 0.06 nmol/L), respectively. Compared with that during saline on day 1, the increment in the plasma pancreatic polypeptide concentration during hypoglycemia was attenuated by ∼50% on day 2 under all three study conditions (P < 0.0001; Fig. 4). The pancreatic polypeptide values during the last hour of hypoglycemia were 373 ± 6 pg/mL (89 ± 1 pmol/L) on day 1 and 216 ± 7 pg/mL (52 ± 2 pmol/L), 214 ± 17 pg/mL (51 ± 4 pmol/L), and 239 ± 10 pg/mL (57 ± 2 pmol/L), respectively, on day 2.

Plasma concentrations of infused insulin were stable and comparable to those during saline on day 1 during all three hyperinsulinemic euglycemic and hypoglycemic clamps on day 2 (Fig. 5). Compared with that during saline on day 1, the increments in the plasma glucagon concentrations during hypoglycemia were reduced under all three conditions on day 2 (P = 0.0263, 0.0008 and <0.0001, respectively; Fig. 5). The glucagon values during the last hour of hypoglycemia during saline infusion on day 1 were 99 ± 1 pg/mL (28 ± 0 pmol/L). The glucagon values during the last hour of hypoglycemia on day 2 were 89 ± 3 pg/mL (26 ± 1 pmol/L), 83 ± 2 pg/mL (24 ± 1 pmol/L), and 86 ± 2 pg/mL (25 ± 1 pmol/L) after saline, adrenergic blockade, and adrenergic plus cholinergic blockade on day 1, respectively.

This was also the case for increments in the plasma cortisol concentrations during hypoglycemia on day 2 (P = 0.0001, <0.0001 and 0.0030) (Fig. 5). The cortisol values during the last hour of hypoglycemia during saline infusion on day 1 were 17.3 ± 0.7 μg/dL (480 ± 19 nmol/L). Cortisol values during the last hour of hypoglycemia on day 2 were 14.2 ± 0.6 μg/dL (390 ± 17 nmol/L), 13.5 ± 1.0 μg/dL (370 ± 28 nmol/L), and 15.0 ± 1.1 μg/dL (410 ± 30 nmol/L) after saline, adrenergic blockade, and adrenergic plus cholinergic blockade on day 1, respectively. Compared with that during saline on day 1, the increment in the plasma growth hormone concentration during hypoglycemia was reduced on day 2 (P = 0.0028; Fig. 5). The growth hormone values during the last hour of hypoglycemia were 12.6 ± 0.5 ng/mL (560 ± 22 pmol/L) on day 1 and 7.7 ± 0.2 ng/mL (340 ± 9 pmol/L) on day 2. In contrast, growth hormone values during the last hour of hypoglycemia on day 2 were not reduced after adrenergic blockade or adrenergic plus cholinergic blockade on day 1 −10.8 ± 0.8 ng/mL (480 ± 35 pmol/L) and 14.0 ± 1.4 ng/mL (620 ± 62 pmol/L), respectively.

We used a well-documented model of HAAF in diabetes (hypoglycemia attenuates the sympathoadrenal response to hypoglycemia the following day) (1–4) and the classic adrenergic and cholinergic antagonists phentolamine, propranolol and atropine (13–15) to test the hypothesis that adrenergic activation, cholinergic activation, or both, mediate the effect of recent antecedent hypoglycemia to reduce the sympathoadrenal response to subsequent hypoglycemia, the key feature of HAAF in diabetes (1–8), in humans.

The data indicate that adrenergic mechanisms mediate the effect of recent antecedent hypoglycemia to reduce the sympathoadrenal response to subsequent hypoglycemia. Morning and afternoon hypoglycemia with saline infusion on day 1 led to attenuated increments in plasma epinephrine and norepinephrine concentrations, markers of the sympathoadrenal, primarily adrenomedullary, response to hypoglycemia (17), during hypoglycemia on day 2. Thus, the phenomenon that we sought to explore mechanistically—recent antecedent hypoglycemia reduces the sympathoadrenal response to subsequent hypoglycemia (1–8)—was reproduced.

However, when adrenergic blockade (without or with cholinergic blockade) was produced during hypoglycemia on day 1, the increments in plasma epinephrine and norepinephrine concentrations during hypoglycemia on day 2 were not attenuated. Thus, combined adrenergic blockade with the nonselective α-adrenergic antagonist phentolamine and the nonselective β-adrenergic antagonist propranolol prevented the effect of hypoglycemia to reduce the sympathoadrenal response to subsequent hypoglycemia. The latter is the key feature of HAAF in diabetes (1–8).

The data do not disclose—but do not categorically exclude—an additional cholinergic mechanism. The sympathoadrenal response to hypoglycemia on day 2 was similar whether the muscarinic cholinergic antagonist atropine was or was not administered with phentolamine and propranolol during hypoglycemia on day 1.

Morning and afternoon hypoglycemia with saline infusion on day 1 also led to attenuated increments in the plasma pancreatic polypeptide concentration, a marker of the parasympathetic response to hypoglycemia (18), during hypoglycemia on day 2. In contrast to the effect on the sympathoadrenal response, adrenergic blockade (without or with cholinergic blockade) during hypoglycemia on day 1 did not prevent the attenuated pancreatic polypeptide response to hypoglycemia on day 2. An altered parasympathetic response is not known to be involved in the pathogenesis of HAAF in diabetes (1).

To the extent the drugs used enter the central nervous system (propranolol and atropine are known to do so), the data do not distinguish between central and peripheral (autonomic) nervous system mediation of the effect of antecedent hypoglycemia to reduce the sympathoadrenal response to subsequent hypoglycemia. However, because the adrenergic antagonists block the heart rate and pressor responses to hypoglycemia, which are normally mediated by the autonomic response, one might speculate that it was blockade of adrenergic sympathoadrenal actions during hypoglycemia on day 1 that prevented an attenuated plasma catecholamine response to hypoglycemia on day 2.

The data confirm substantially higher plasma epinephrine and norepinephrine concentrations in response to hypoglycemia during adrenergic blockade on day 1 (15). That is expected because both catecholamines are cleared through β-adrenergic mechanisms in humans (19). (Given the short half-times of the antagonists and the catecholamines, a carry-over effect to the next day is unlikely.) The current findings of higher plasma concentrations of infused insulin and of endogenous growth hormone and cortisol during hypoglycemia during adrenergic blockade suggest that adrenergic mechanisms are also involved in the clearance of these hormones. The growth hormone responses to hypoglycemia, like the epinephrine and norepinephrine responses, were reduced after hypoglycemia with saline infusion but not after hypoglycemia with adrenergic blockade (without or with cholinergic blockade). An interesting finding was that the pancreatic polypeptide, glucagon, and cortisol responses to hypoglycemia were reduced on day 2 regardless of the day 1 conditions. Among these, a reduced glucagon response is a feature of HAAF (1). Thus, the data further support the view that the mechanisms of the reduced glucagon and epinephrine responses to hypoglycemia in HAAF are different (1).

In conclusion, the data indicate that adrenergic mechanisms mediate the effect of recent antecedent hypoglycemia to reduce the sympathoadrenal response to subsequent hypoglycemia, the key feature of HAAF in diabetes, in humans.

This study was partly supported by National Institutes of Health Grants R37-DK-27085 and UL1-RR-24992 and by a fellowship award from the American Diabetes Association.

P.E.C. has served as a consultant to Bristol-Myers Squibb/AstraZeneca, MannKind Corp., Merck & Co., and Novo Nordisk in the past year. No other potential conflicts of interest relevant to this article were reported.

R.R. and P.E.C. designed the study, R.R. conducted it, and R.R. and P.E.C. analyzed the data and wrote the manuscript.

The authors acknowledge the assistance of the following Washington University staff: Licia Rowe, Laura Karsteter, Nilima Parikh, Tanya Eden, Shirley Frei, Janice Bathon, Paula Blood, and David Gibson of the Clinical Research Unit and of the Core Laboratory for Clinical Studies; the technical assistance of Krishan Jethi, MS; and the statistical assistance of Ling Chen, PhD, MPH. Janet Dedeke prepared this manuscript.