To establish the antihyperglycemic mechanisms of metformin in non-insulin-dependent diabetes mellitus (NIDDM) independently of the long-term, aspecific effects of removal of glucotoxicity, 21 NIDDM subjects (14 obese, 7 nonobese) were studied on two separate occasions, with an isoglycemic (plasma glucose approximately 9 mM) hyperinsulinemic (two-step insulin infusion, 2 h each, at the rate of 4 and 40 mU.m-2.min-1) clamp combined with [3-3H]glucose infusion and indirect calorimetry, after administration of either metformin (500 mg per os, at -5 and -1 h before the clamp) or placebo. Compared with placebo, hepatic glucose production (HGP) decreased approximately 30% more after metformin (from 469 +/- 50 to 330 +/- 54 mumol/min), but glucose uptake did not increase. Metformin suppressed free fatty acids (FFAs) by approximately 17% (from 0.42 +/- 0.04 to 0.35 +/- 0.04 mM) and lipid oxidation by approximately 25% (from 4.5 +/- 0.4 to 3.4 +/- 0.4 mumol.kg-1.min-1) and increased glucose oxidation by approximately 16% (from 16.2 +/- 1.4 to 19.3 +/- 1.3 mumol.kg-1.min-1) compared with placebo (P < 0.05), but did not affect nonoxidative glucose metabolism, protein oxidation, or total energy expenditure. Suppression of FFA and lipid oxidation after metformin correlated with suppression of HGP (r = 0.70 and r = 0.51, P < 0.001). The effects of metformin in obese and nonobese subjects were no different. We conclude that the specific, antihyperglycemic effects of metformin in the clinical condition of hyperglycemia in NIDDM are primarily due to suppression of HGP, not stimulation of glucose uptake, and are mediated, at least in part, by suppression of FFA and lipid oxidation.

To test the hypothesis that cortisol secretion plays a counterregulatory role in hypoglycemia in humans, four studies were performed in eight normal subjects. In all studies, insulin (15 mU.m-2.min-1) was infused subcutaneously (plasma insulin 27 +/- 1 microU/ml). In study 1, plasma glucose concentration and glucose fluxes [( 3-3H]glucose), substrate, and counterregulatory hormone concentrations were simply monitored, and plasma glucose decreased from 89 +/- 2 to 52 +/- 2 mg/dl for 12 h. In study 2, (pituitary-adrenal-pancreatic clamp), insulin and counterregulatory hormone secretion (except for catecholamines) was prevented by somatostatin (0.5 mg/h, iv) and metyrapone (0.5 g/4 h, per os), and glucagon, cortisol, and growth hormone were infused to reproduce the concentrations of study 1. In study 3 (lack of cortisol increase), the pituitary-adrenal-pancreatic clamp was performed with maintenance of plasma cortisol at basal levels, and glucose was infused, whenever needed, to reproduce plasma glucose concentration of study 2. Study 4 was identical to study 3, but exogenous glucose was not infused. Isolated lack of cortisol increase caused a approximately 22% decrease in hepatic glucose production (P less than 0.01) and a approximately 15% increase in peripheral glucose utilization (P less than 0.01), which resulted in greater hypoglycemia (37 +/- 2 vs. 52 +/- 2 mg/dl, P less than 0.01) despite compensatory increases in plasma epinephrine. Lack of cortisol response also reduced plasma free fatty acid, beta-hydroxybutyrate, and glycerol concentrations approximately 50%. We conclude that cortisol normally plays an important counterregulatory role during hypoglycemia by augmenting glucose production, decreasing glucose utilization, and accelerating lipolysis.

To establish the glycemic threshold for onset of neuroglycopenia (impaired cognitive function, measured by the latency of the P300 wave), activation of hormonal counterregulation and hypoglycemic symptoms, 12 normal subjects were studied either under conditions of insulin-induced, glucose-controlled plasma glucose decrements, or during maintenance of euglycemia. A decrement in plasma glucose concentration from 88±3 to 80±1 mg/dl for 150 min did not result in changes in the latency of the P300 wave nor in an activation of counterregulatory hormonal response. In contrast, a greater decrement in plasma glucose concentration from 87±3 to 72±1 mg/dl for 120 min caused an increase in the latency of the P300 wave (from 301±12 to 348±20 ms, P < 0.01), a subsequent increase in all counterregulatory hormones but no hypoglycemic symptoms. Finally, when plasma glucose concentration was decreased in a stepwise manner from 88±2 to 50±1 mg/dl within 75 min, the increase in the latency of the P300 wave was correlated with the corresponding plasma glucose concentration (r = -0.76, P < 0.001). The glycentic threshold for hypoglycemic symptoms was 49±2 mg/dl.Thus, in normal man the glycemic threshold for neuroglycopenia (72±1 mg/dl) is greater than currently thought; the hormonal counterregulation follows the onset of neuroglycopenia; the hypoglycemic symptoms are a late indicator of advanced neuroglycopenia.

To test the hypothesis that growth hormone secretion plays a counterregulatory role in prolonged hypoglycemia in humans, four studies were performed in nine normal subjects. Insulin (15 mU.M-2.min-1) was infused subcutaneously (plasma insulin 27 +/- 2 microU/ml), and plasma glucose decreased from 88 +/- 2 to 53 +/- 1 mg/dl for 12 h. In study 1, plasma glucose, glucose fluxes (D-[3-3H]glucose), substrate, and counterregulatory hormone concentrations were simply monitored. In study 2 (pituitary-adrenal-pancreatic clamp), insulin and counterregulatory hormone secretions (except for catecholamines) were prevented by somatostatin (0.5 mg/h iv) and metyrapone (0.5 g/4 h po), and glucagon, cortisol, and growth hormone were reinfused to reproduce the concentrations of study 1. In study 3 (lack of growth hormone increase), the pituitary-adrenal-pancreatic clamp was performed with maintenance of plasma growth hormone at basal levels, and glucose was infused whenever needed to reproduce plasma glucose concentration of study 2. Study 4 was identical to study 3, but exogenous glucose was not infused. Isolated lack of a growth hormone response caused a decrease in hepatic glucose production and an increase in glucose utilization that resulted in an approximately 25% greater hypoglycemia despite compensatory increases in plasma catecholamines. Plasma free fatty acid, 3-beta-hydroxybutyrate, and glycerol concentrations were reduced approximately 50%. It is concluded that growth hormone normally plays an important counterregulatory role during hypoglycemia by augmenting glucose production, decreasing glucose utilization, and accelerating lipolysis.

To ascertain whether the dawn phenomenon occurs in nondiabetic individuals and, if so, whether it is due to an increase in glucose production or a decrease in glucose utilization, we determined plasma concentrations of glucose, insulin, C-peptide, and counterregulatory hormones, as well as rates of glucose production, glucose utilization, and insulin secretion at one-half-hourly intervals between 1:00 and 9:00 a.m. in eight normal volunteers. After 5:30 a.m., plasma glucose, insulin, and C-peptide concentrations all increased significantly; rates of glucose production, glucose utilization, and insulin secretion also increased (all P less than 0.05). Plasma cortisol, epinephrine, and norepinephrine increased significantly from nocturnal nadirs between 4:00 and 6:30 a.m. Plasma growth hormone, which had increased episodically between 1:00 and 4:30 a.m., decreased thereafter nearly 50% (P less than 0.05). Plasma glucagon did not change significantly throughout the period of observation. These results indicate that a dawn-like phenomenon, initiated by an increase in glucose production, occurs in nondiabetic individuals. Thus, early morning increases in plasma glucose concentrations and insulin requirements observed in IDDM and NIDDM may be an exaggeration of a physiologic circadian variation in hepatic insulin sensitivity induced by antecedent changes in catecholamine and/or growth hormone secretion.