Abstract

This article examines the role of the autonomic nervous system in mediating the increase of glucagon secretion observed during insulin-induced hypoglycemia (IIH). In the first section, we briefly review the importance of the a-cell response in recovery from hypoglycemia under both physiologic conditions and pathophysiologic conditions, such as type 1 diabetes. We outline three possible mechanisms that may contribute t o increased glucagon secretion during hypoglycemia but emphasize autonomic mediation. In the second section, we review the critical experimental data in animals, nonhuman primates, and humans suggesting that, in the absence of diabetes, the majority of the glucagon response to IIH is mediated by redundant autonomic stimulation of the islet a-cell. Because the glucagon response t o hypoglycemia is often impaired in patients with type 1 diabetes, in the third section, we examine the possibility that autonomic impairment contributes t o the impairment of the glucagon response in these patients. We review two different types of autonomic impairment. The first is a slow-onset and progressive neuropathy that worsens with duration of diabetes, and the second is a rapid-onset, but reversible, autonomic dysfunction that i s acutely induced by antecedent hypoglycemia. We propose that both types of autonomic dysfunction can contribute t o the impaired glucagon responses in patients with type 1 diabetes. In the fourth section, we relate restoration of these glucagon responses t o restoration of the autonomic responses t o hypoglycemia. Finally, in the fifth section, we summarize the concepts underlying the autonomic hypothesis, the evidence for it, and the implications of the autonomic hypothesis for the treatment of type 1 diabetes. Diabetes 47:995-1005, 1998 From the Division of Endocrinology, Metabolism and Nutrition (G.J.T.), Department of Medicine, University of Washington; Veterans Affairs Puget Sound Health Care System (G.J.T.), Seattle, Washington: the Department of Medicine (B.A.), Lund University, Malmo University Hospital, Malmo, Sweden; and the Department of Nutrition (P.J.H.), University of California, Davis, California. Address correspondence and reprint requests to Gerald J. Taborsky Jr, PhD, Division of Endocrinology and Metabolism (151), Veterans Affairs Puget Sound Health Care System, 1660 South Columbian Way, Seattle, WA 98108. E-mail: taborsky@u.washington.edu. Received for publication 2 February 1998 and accepted in revised form 21 April 1998. DAN, diabetic autonomic neuropathy; IIH, insulin-induced hypoglycemia; PP, pancreatic polypeptide; VIP, vasoactive intestinal peptide. GLUCAGON SECRETION DURING HYPOGLYCEMIA: CLINICAL IMPORTANCE AND POTENTIAL MECHANISMS Clinical importance. Hypoglycemia is a major acute complication of type 1 diabetes, contributing to both its morbidity and its mortality (1). Attempts to reduce the long-term complications associated with type 1 diabetes, such as retinopathy and nephropathy, by reducing hyperglycerniavia intensive insulin therapy have unfortunately exacerbated this acute complication: intensive insulin therapy produces a threefold increase of the incidence of severe hypoglycemia (2,3). For this reason, hypoglycemia is a major limiting factor in improving glycemic control (4) and, therefore, in preventing or reducing the long-term complications associated with type 1 diabetes. Consequently, it is important to understand the physiologic mechanisms that limit the severity and duration of hypoglycemic episodes. The work of Gerich and Cryer (5,6) established the importance of glucagon and epinephrine in limiting the glucose nadir and promoting rapid recovery of plasma glucose from rapid and transient hypoglycemia in nondiabetic human subjects. Later studies, primarily by Bolli et al., also suggested important roles for glucagon (7) and epinephrine (8) in countering the slow-onset, but progressive, hypoglycemia that is more often encountered clinically (9). Because most patients who have had type 1 diabetes for 1-2 decades have totally lost their glucagon response to IIH and have blunted epinephrine responses, these studies helped to explain the impairn~ent of glucose recovery observed in patients with long-term type 1 diabetes NON-DIABETIC Potential mechanisms. Three different mechanisms have been proposed to mediate the a-cell response to hypoglycemia A second mechanism that could potentially mediate increased glucagon secretion in response to hypoglycemia involves a local inhibitory effect of endogenous insulin on neighboring acells A thrd potential mediator of the glucagon response to hypoglycemia is the autonomic input to the a-cell (33) All three of these autonomic inputs to the a-cell-sympathetic nerves, parasympathetic nerves, and adrenal medullaare activated by hypoglycemia (33). Cannon first demon-DIABETES, VOL. 47, JULY 1998 m . m m . strated over 70 years ago that hypoglycemia stimulates the secretion of epinephrine from the adrenal medulla. In contrast, the activation of pancreatic sympathetic nerves by hypoglycemia has only recently been demonstrated Each of these three autonomic inputs to the a-cell is capable of stimulating glucagon secretion (33). Epinephrine stimulates glucagon secretion both in vitro (38) and in vivo (39). Electrical stimulation of pancreatic sympathetic nerves (40) or local infusion of the classical sympathetic neurotransmitter norepinephrine (41) or a pancreatic sympathetic neuropeptide, such as galanin Although the three mechanisms above have been discussed separately, there is evidence for autonomic participation both in the effect of glucose on the a-cell in vitro and in the suppression of insulin secretion in vivo. For example, in vitro perfusion of the pancreas with low glucose media can release norepinephrine from the sympathetic nerve terminals of the pancreas (48), and a-adrenergic blockade during such perfusions nearly abolishes the glucagon response to glucopenia (49). Thus, surprisingly, there is evidence for autonomic mediation of glucagon response to glucopenia, even in vitro. These findings provide an alternative to the interpretation that hypoglycemia directly stimulates the a-cell. Evidence in vivo suggests that part of the suppression of endogenous insulin secretion seen during IIH is due to activation of the autonomic nervous system. The evidence for the effect of hyperinsulinemia on endogenous insulin secretion includes suppression of C-peptide secretion during hyperinsulinemic euglycenuc clamps (32) AUTONOMIC CONTRIBUTION IN NONDIABETIC ANIMALS AND HUMANS Animal studies. In 1989, when we first reviewed the literature on autonomic mediation of glucagon secretion during hypoglycemia (33), the mqjority of studies in humans were negative. Fhthermore, only two studies conducted in animals, one in rats (53) and on(? in calves (54), suggested a sigruficant autonomic contributi\ .n to hypoglycemia-induced glucagon secretion. We hypothesized that redundant stimulation of glucagon secretion by the parasympathetic and sympathoadrenal divisions of the autonomic nervous system Several other studies have provided evidence of central neural mediation. For example, preventing central, but not peripheral, hypoglycemia in dogs with selective carotid and vertebral artery glucose infusions markedly attenuates both parasympathetic (PP) and sympathoadrenal (epinephrine) responses to IIH and eliminates the increase of glucagon seen when the brain and the pancreas are simultaneously exposed to hypoglycemia (59). Further, lesioning of the ventromedial hypothalamus in rats (60) or microdialysis of glucose into this region (61) significantly impairs both autonomic activation and the glucagon responses to IM. These studies demonstrate the necessity of neural activation for the glucagon response to hypoglycemia in animals. Other studies in which autonomic inputs were sequentially blocked support the hypothesized redundancy of autonomic stimulation of glucagon secretion during hypoglycemia. For example, the increase of plasma glucagon during IIH in rats was not significantly affected by parasympathetic cholinergic blockade with methylatropine or by a-and Padrenergic blockade with tolazoline and propranolol, but the combination of cholinergic and adrenergic blockade markedly impaired the glucagon response (62). There is additional evidence for at least partial redundancy between parasympathetic and sympathoadrenal mechanisms during marked hypoglycemia in mice (56) and calves (54). More recently, in dogs, we found that either the direct sympathetic innervation of the pancreas (63) or adrenal medullary activation (P.J.H., T.O. Mundinger, G.J.T., unpublished observations) can mediate an intact glucagon response to marked hypoglycemia independently of the vagal parasympathetic input. Therefore, a significant autonomic contribution to increased glucagon secretion during hypoglycemia has been documented in nondiabetic animals, and the response appears to be redundantly mediated by the three autonomic inputs to the pancreas However, it was not known whether different mechanisms were operating to regulate glucagon secretion during hypoglycemia in humans and other primates or whether the negative results obtained from the earlier human studies were the result of an unrecognized redundancy similar to that demonstrated in animals. Evidence for the latter interpretation is provided by studies in rhesus monkeys, in which the ganglionic blocking agent trimethaphan was used to impair autonomic activation, while allowing systemic hypoglycemia and hyperinsulinemia. Trimethaphan infusion nearly abolished the autonomic (epinephrine, norepinephrine, and PP) responses and reduced the glucagon response to hypoglycemia of 35 mgldl(1.9 mrnol~l) by -75% (57), indicating that autonomic activation mediates a substantial portion of the glucagon response in a primate species. In the same study, high doses of atropine combined with a-and P-adrenergic blockade produced a similar degree of impairment of the glucagon response. Thus, the glucagon response to hypoglycemia in nonhuman primates is mediated via activation of the autonomic nervous system, as it is in other animal species. Human studies. Indirect support for autonomic mediation of the glucagon response in humans has been provided by studies of the acute autonomic dysfunction induced by prior episodes of hypoglycemia The impairment of autonomic activation after recent hypoglycemia appears to be mediated by the increase of circulating cortisol evoked by antecedent episodes of hypoglycemia. Cortisol, in turn, may impair autonomic activation (67) by producing a feedback inhibition of the central corticotropinreleasing hormone (68) thought to be involved in mediating sp~pathoadrenal activation during hypoglycemia (69). Evidence for the role of cortisol includes induction of autonomic and glucagon impairment by infusion of cortisol at rates that mimic the plasma cortisol response to hypoglycemia (70) and prevention of impaired autonomic and glucagon responses by blocking the cortisol response to antecedent hypoglycemia with metyrapone, an inhibitor of cortisol synthesis (71). Up-regulation of brain glucose transport may be a separate mechanism that reduces autonomic responses to hypoglycemia. This mechanism is likely to be prominent after long repeated exposure to hypoglycemia, since those are the conditions in which it has been demonstrated both in animals (72,73) and in humans (74). Hypoglycemia-induced cortisol secretion may also be involved in this mechanism of autonomic impairment, since cortisol can induce up-regulation of brain glucose transporters in animals (75). It should be pointed out that up-regulation of brain glucose transport would lessen central glucose deprivation, the stimulus for autonomic activation, whereas the acute effects of cortisol, mediated via suppression of central corticotropin-releasing hormone, would inhibit autonomic outflow even in the presence of central glucopenia. Additional indirect support for autonomic mediation in humans is provided by studies in patients with chronic autonomic dysfunction resulting from Chagas's disease or ShyDrager syndrome. These patients have deficient parasympathetic (PP) (76), adrenal medullary (epinephrine) (77), and glucagon responses (76,77) to hypoglycemia. In contrast, there are two studies that found that classical cholinergic and adrenergic receptor antagonists did not reduce the glucagon response to hypoglycemia in nondiabetic human subjects An alternative explanation is that the lower doses of autonomic blocking agents used in studies in humans (8,79) are not sufficient to totally block the effects of the reflex autonomic activation produced by hypoglycemia For example, the dose of atropine typically used in human studies (1 mg) does not block a cardiovascular parasympathetic reflex (Valsalva); a fourfold higher dose is needed to suppress this response (82). Thus, studies using different doses of autonomic blockers might resolve these discrepancies between human and animal studies. Given the failure of adrenergic and muscarinic blockade to reduce the glucagon response to hypoglycemia in humans, other studies specifically designed to address the role of the autonomic nervous system in nondiabetic humans were necessary. In a recent study, the ganglionic blocker trimethaphan was infused to impair autonomic activation in nondiabetic human subjects. Trimethaphan impaired the increases of plasma PP, epinephrine, and norepinephrine during hypoglycemia of 45 mgldl by 7040% and reduced the glucagon response by 75% GLUCAGON RESPONSE TO HYPOGLYCEMIA IN TYPE 1 DIABETES Impaired a-cell responses. The glucagon response to IIH in patients with type 1 diabetes can be normal, impaired, or totally absent. Before investigators started to examine the effects of intensive insulin therapy, the available data suggested that there was a progressive loss of this response related to the duration of type 1 diabetes. For example, Bolli et al. After clinical trials of intensive insulin therapy and the associated increased incidence of hypoglycemia (2,3), many reports of impaired glucose counterregulation appeared (10-13), even in patients with diabetes of short duration. This impairment of glucose recovery was related to markedly impaired glucagon (1 1,13) and epinephrine (12,13) responses to hypoglycemia. There were also reports of a very early loss of the glucagon response to hypoglycemia in newly diagnosed diabetic children (85,86). Reports of severe and pro-DIABETES, VOL. 47, JULY 1998 longed nocturnal hypoglycemia in children (87) and of a threefold increase in the incidence of hypoglycen~ia associated with intensive insulin therapy in adults (2,3) suggested a contribution of prior hypoglycemia to this early loss of acell responses. A potential mechanism involving acute autonomic dysfunction is discussed below. Mechanisms Hgpoglgcemiaper se. The mechanism for the slow and p r e gressive loss of the glucagon response to hypoglycemia in type 1 patients has been a subject of controversy. The loss is selective, i.e., patients with long-standing type 1 diabetes have essentially no glucagon response to IIH yet have a normal glucagon response to arginine (84). Thus, the secretory capacity of the a-cell is normal, yet the a-cell is either unresponsive to the stimuli specifically associated with IIH (see below) or these stimuli are reduced in type 1 diabetes. As p r e viously discussed, hypoglycemia per se, suppression of endogenous insulin secretion, and activation of the autonomic nervous system have all been proposed as mediators of the stimulation of the glucagon response to hypoglycemia. It is possible that an intrinsic defect in the a-cell's response to direct hypoglycemic stimulation is responsible for loss of this a-cell response in type 1 diabetes, as originally proposed by Gerich et al. (&I). Restoration of euglycemia in diabetic rats does improve their glucagon responses to hypoglycemia (88) consistent with a partially reversible defect in the a-cell's recognition of glucose that was induced by the chronic hyperglycemia of diabetes. However, restoration of euglycemia has also been shown to improve both residual P-cell function Some have even proposed that the progressive loss of the a-cell response to hypoglycemia in patients with type 1 diabetes is due to progressive loss of their residual P-cell function (93). This hypothesis implies that many newly diagnosed type 1 diabetic patients retain a small amount of P-cell function, as has been documented both by residual insulin staining in one-third of the islets of autopsied pancreases More subtle, subclinical forms of neuropathy may be pres ent much earlier in the natural history of type 1 diabetes. These may be detected with more sensitive autonomic tests, such as measurement of heart rate variability (98), scintography of the heart (99), and muscle sympathetic nerve activity (100). Indeed, one-half to three-fourths of patients with type 1 diabetes who did not exhibit the usual signs of clinical DAN had impaired autonomic function, as assessed by these more sensitive tests (99-101). Other studies report the presence of subclinical autonomic neuropathy in one-half of all patients with type 1 diabetes (97). Some of this autonomic dysfunction, however, may be unrelated to the autonomic inputs to the acell. Therefore, it is important to determine if the auto nomic responses to IIH per se are impaired in patients with type 1 diabetes before the development of clinical DAN. Results from a recent study have demonstrated just that: the epinephrine and the PP responses to IIH are impaired in patients that have type 1 diabetes without clinical signs of DAN (102). Finally, it is important to recognize that the response of the sympathetic nerves that innervate the a-cell DIABETES, VOL. 47, JULY 1998