Epinephrine Secretion, Hypoglycemia Unawareness, and Diabetic Autonomic Neuropathy

Is the Failure of Acute Glucose Counterregulation the Result of Autonomic Neuropathy?

Autonomic neuropathy does not adequately explain the failure of acute glucose counter-regulation in patients with type I diabetes, because some patients with inadequate epinephrine responses to hypoglycemia have no signs, symptoms, or cardiovascular reflex abnormalities indicative of autonomic dysfunction [2, 4-9]. Failure of glucose counterregulation in some of these patients may be the result of tight antecedent control or a recent hypoglycemic event. Even healthy persons have a decreased symptomatic response as well as a decreased epinephrine response to hypoglycemia 16 hours after a standardized hypoglycemic stimulus Figure 1[10, 11]. This proves that glucose counterregulation can be affected independent of autonomic neuropathy. In addition, hyperinsulinemia itself has recently been shown to suppress the counterregulatory response to hypoglycemia [12]. These observations indicate that the interpretation of counterregulatory response to hypoglycemia is more complex than previously realized. However, we disagree with the recent suggestion [13] that failure in glucose counterregulation in diabetes is solely the result of metabolic abnormalities associated with diabetes and that preexisting autonomic neuropathy has no effect on the adrenergic response to hypoglycemia. Polinsky and colleagues [14] first showed that patients with idiopathic autonomic neuropathy fail to secrete epinephrine in response to hypoglycemia; this was confirmed in patients with both pure autonomic failure and multiple system atrophy. Because these patients did not have diabetes and presumably never had a spontaneous hypoglycemic event or hyperinsulinemia, these studies prove that autonomic neuropathy can disrupt the adrenergic response to hypoglycemia.

The conflicting data on the relation between glucose counterregulation and autonomic neuropathy in patients with diabetes are the result of the different criteria that have been used in the published studies for the diagnosis of autonomic neuropathy. With the continuing emergence of new autonomic function tests, divergent and often poorly validated diagnostic criteria have been proposed [6]. In one study [7] that disputed the association between autonomic neuropathy and counterregulatory hormone secretion, the criteria for diagnosing autonomic neuropathy were not even stated.

Confusion about the definition and classification of diabetic polyneuropathy led to the organization of a conference of experts who convened in San Antonio in 1988 and who developed standardized terminology and staging criteria for patients with diabetic polyneuropathy [15]. The major stages defined by the San Antonio committee included: N0, no neuropathy; N1, asymptomatic neuropathy; N2, symptomatic neuropathy; and N3, disabling polyneuropathy. One of the limitations of this approach is that the development of even one sign or symptom of neuropathy places the patient in stage N2 (symptomatic neuropathy). Some of the symptoms, such as pain or erectile dysfunction, are difficult to evaluate objectively. Thus, despite rigorous efforts at classification, patients who fit the clinical criteria for stage N2 (symptomatic neuropathy) remain heterogenous. For this reason, attempts to test the relation between autonomic neuropathy and counterregulatory responses have yielded conflicting results. We diagnosed autonomic neuropathy only if patients had clinically significant orthostatic hypotension, a decrease in systolic blood pressure of more than 20 mm Hg after shifting from the supine to the upright posture [3]. If orthostatic hypotension is unaccompanied by a compensatory tachycardia and is chronically present in a euvolemic patient who is not taking vasodilators or other drugs that might decrease blood pressure, the presence of autonomic neuropathy is unequivocal; this provides a reference standard against which new tests of autonomic function can be evaluated. Many diabetic patients with orthostatic hypotension have multiple features of autonomic neuropathy, and their disease is similar in its severity to that of patients with idiopathic autonomic neuropathy in whom the failure of the epinephrine response to hypoglycemia was first documented [14]. Five independent groups of investigators have confirmed our finding [3] that diabetic patients with overt autonomic neuropathy have diminished epinephrine responses to hypoglycemia [16-20]. Moreover, degenerative changes in the cholinergic neurons innervating the adrenal medulla have been documented in diabetic rats and have been shown to attenuate ex vivo epinephrine secretion in response to chemical and electrical stimulation [21]. Those clinical studies that failed to confirm a relation between counterregulatory hormone secretion and autonomic neuropathy defined the latter on the basis of tests of heart rate variability [22] or sudomotor reflex testing [6]. This approach has been rationalized by invoking the simplistic view and misconception that autonomic neuropathy, if diagnosed anywhere, must be present diffusely [6]; however, no data support this concept. Moreover, no evidence exists that patients with abnormalities on selected autonomic function tests (isolated abnormalities in sudomotor or cardiovascular autonomic function) subsequently develop diffuse autonomic neuropathy. In practice, tests of heart rate variability and sudomotor function diagnose autonomic neuropathy that is much less severe than is typically present in patients with orthostatic hypotension. In fact, many patients who do poorly in tests of heart rate variability have no signs or symptoms of autonomic neuropathy and, thus, have subclinical disease, stage N1 by the San Antonio classification [15].

The importance of subclinical autonomic neuropathy has been questioned because progression to overt autonomic neuropathy (from N1 to N2) is unusual. Sampson and colleagues [23] reported that none of 24 patients with subclinical autonomic neuropathy progressed to overt neuropathy even after a 10-year follow-up period. It is easy to envision how subclinical neuropathy (N1) (which is very common, occurring in half of all patients with type I diabetes, and which may cause only subtle abnormalities on standardized testing) would affect counterregulatory hormone secretion less than symptomatic (N2) or disabling (N3) neuropathy, which signify diffuse or advanced disease. Webb and Castaner [20] made direct comparisons of counterregulatory hormone secretion in patients with subclinical autonomic neuropathy (as diagnosed from tests of parasympathetic cardiac innervation) and severe sympathetic neuropathy (as evidenced by orthostatic hypotension) and showed that only the latter group failed to secrete epinephrine in response to hypoglycemia (Figure 2). Since orthostatic hypotension is a relatively uncommon complication of diabetic autonomic neuropathy, the latter explains failure of glucose counterregulation in only a minority of patients with inadequate epinephrine responses to hypoglycemia.

The lack of correlation between counterregulatory hormone secretion and most autonomic function tests have led some investigators to conclude that the presence of autonomic neuropathy does not affect a diabetic patient's ability to secrete counterregulatory hormones [6, 13, 22]. An alternative explanation is that the autonomic function tests most often used in the diagnosis of autonomic neuropathy, especially heart rate variability with deep breathing, fail to test the specific central and autonomic mechanisms that mediate counterregulatory hormone secretion. An analogy with the evaluation of patients with impotence may be appropriate. Many patients with erectile dysfunction have normal vasomotor reflex test results. Does it follow that impotence is not a result of autonomic neuropathy? Campbell and colleagues [24] recognized that autonomic failure in patients with diabetes may involve certain autonomic functions yet spare others and proposed the term “selective autonomic neuropathy” to describe the failure of counterregulatory hormone secretion in patients with diabetes. Evidence supporting this concept was provided by Kennedy and colleagues [25], who observed a correlation between pancreatic polypeptide and epinephrine secretion after hypoglycemia in patients with type I diabetes. The pancreatic polypeptide response to hypoglycemia is a function of the parasympathetic innervation of the pancreas, and patients with overt diabetic autonomic neuropathy typically fail to secrete this polypeptide after hypoglycemia. The patients described by Kennedy and colleagues [25], however, had normal heart rate variability with deep breathing and, therefore, did not have even subclinical autonomic neuropathy by the usual criteria. Thus, their diminished secretion of pancreatic polypeptide and epinephrine may have been the result of a variant of autonomic neuropathy that selectively affects counterregulatory hormone secretion but spares cardiac innervation and vasomotor reflexes. In some patients, selective autonomic neuropathy has evolved into a more generalized disease with typical involvement of cardiovascular reflexes [26].

In summary, failure of acute glucose counterregulation in insulin-dependent diabetes can be attributed to overt autonomic neuropathy in only a few patients. In the remainder, decreased epinephrine secretion is the result of either failure of the central nervous system to recognize hypoglycemia or the consequence of a selective autonomic neuropathy that disrupts counterregulatory hormone secretion but spares cardiovascular reflexes.

Does Autonomic Neuropathy Affect Hypoglycemia Awareness by Mechanisms Independent of Counterregulatory Hormone Secretion?

Although subclinical autonomic neuropathy does not necessarily lead to failure of counterregulatory hormone secretion, it may place patients at risk for severe hypoglycemia by other mechanisms. It is well known that some of the symptoms of hypoglycemia (tremor, diaphoresis, palpitations) are mediated by autonomic nervous system activation, which presumably can take place in the absence of changes in circulating catecholamine concentrations. Hepburn and colleagues [4] observed that 67% of patients with hypoglycemia unawareness did poorly on autonomic function tests; in contrast, only 37% of patients with normal awareness of hypoglycemia had evidence of subclinical autonomic neuropathy. Similarly, Barkai and colleagues [5] observed that 58% (7 of 12) of diabetic children with a history of severe hypoglycemia had subclinical autonomic neuropathy (poor performance on three or more vasomotor reflex tests); in contrast, no child (0 of 57) without a history of severe hypoglycemia had subclinical autonomic neuropathy by this criterion. These studies, taken together, show that subclinical autonomic neuropathy, as diagnosed from measurements of heart rate variability, may be associated with an increased risk for hypoglycemia unawareness. Ryder and colleagues [6], however, who diagnosed subclinical neuropathy with sudomotor function tests, failed to confirm a relation between autonomic function and propensity to hypoglycemia.

Does the Epinephrine Response to Experimental Hypoglycemia Predict a Patient's Risk for Developing Hypoglycemia Clinically?

The physiologic importance of epinephrine secretion and the interpretation of plasma epinephrine concentrations in hypoglycemic patients have also been topics of recent debate. Most [3, 27-30], but not all [19], studies confirm that a correlation exists between plasma epinephrine concentrations and adrenergic warning symptoms in patients with experimental hypoglycemia, and it is generally believed that an inability to secrete epinephrine places a diabetic patient at risk for developing severe hypoglycemia. Some studies [8], however, have indicated that little correlation exists between a patient's epinephrine response to experimental hypoglycemia and his or her actual risk for developing hypoglycemia clinically. This suggests that other mechanisms may cause hypoglycemia unawareness in some patients. We have encountered two patients with hypoglycemia unawareness and with recurrent hypoglycemic coma who had normal or even excessive epinephrine responses to hypoglycemia. Accordingly, decreased cardiac β-adrenergic receptors have been shown in rats with streptozotocin-induced diabetes [31], and decreased chronotropic responses to isoproterenol have been described in diabetic patients with hypoglycemia unawareness [32]. Hilsted and colleagues [33], however, found that most patients with insulin-dependent diabetes mellitus have normal hemodynamic and metabolic responses to the infusion of epinephrine; those patients with autonomic neuropathy were excessively sensitive, presumably because of denervation hypersensitivity. Thus, the physiologic significance of plasma epinephrine concentrations may be different in patients who present with hypoglycemia unawareness, because they may have decreased adrenergic sensitivity, whereas those patients who present with symptomatic autonomic neuropathy may have increased adrenergic sensitivity.

The glycemic threshold at which epinephrine secretion is initiated may be more important in determining the risk for hypoglycemia than is the peak epinephrine response. Simonson and colleagues [34] were the first to show that intensive insulin therapy places patients at risk for hypoglycemia primarily by altering the glycemic threshold at which symptomatic responses and counter-regulatory hormone secretion is initiated (Figure 3). Not surprisingly, severe hypoglycemia was three times more common among the intensely treated participants in the Diabetes Control and Complications Trial than in the conventionally treated patients [35]. Grimaldi and colleagues [7] found that even conventionally treated patients with hypoglycemia unawareness have an increased glycemic threshold for counterregulatory hormone secretion; a lower blood sugar is needed to initiate a response. These and other studies reviewed elsewhere [36] indicate that failure of the central nervous system glucosensors to detect a decreased blood sugar is probably a common cause of failure of glucose counterregulation in patients with diabetes. Dysfunction of the central glucose sensor typically results from hypoglycemia itself. Thus, meticulous avoidance of even mild hypoglycemia, by frequent monitoring of plasma glucose levels, may be the most practical means for preventing the failure of glucose counterregulation and preventing its consequence, hypoglycemia unawareness.

Summary and Conclusions

The failure of some type I diabetic patients to secrete epinephrine and glucagon in response to hypoglycemia has been documented by many investigators, and most studies have confirmed that an inability to secrete these counterregulatory hormones places patients at risk for developing clinical hypoglycemia [27-30]. Inadequate acute glucose counterregulation can result from multiple mechanisms. Failure of central glucoreceptors to recognize hypoglycemia and to activate counterregulation may be the most common [36]. Decreased central recognition of hypoglycemia results from either strict antecedent glucose control [34] or from a recent hypoglycemic event [10, 11, 13]. Controversy about the relation between autonomic neuropathy and counterregulatory hormone secretion has arisen because divergent criteria have been used in the published studies for the diagnosis of autonomic neuropathy. Advanced adrenergic neuropathy, as evidenced by orthostatic hypotension, generally leads to decreased epinephrine secretion after hypoglycemia [3, 16-20]. Subclinical neuropathy, however, as diagnosed from measurement of heart rate variability, may diminish the awareness of hypoglycemia [4, 5] but does not affect counterregulatory hormone secretion [20, 22, 30]. Failure of counterregulatory hormone secretion in some patients with type I diabetes, however, may represent a selective autonomic neuropathy; the disease has limited the patient's ability to secrete epinephrine and pancreatic polypeptide in response to hypoglycemia even though it has spared the autonomic neurons responsible for cardiovascular reflexes [25]. Finally, recent provocative reports [32] indicate that decreased responsiveness to adrenergic stimuli may cause hypoglycemia unawareness in some patients. Further documentation of this mechanism is required, and its relative importance with respect to other mechanisms needs to be established. These questions are increasingly important clinically because the Diabetes Control and Complications Trial [1, 35] has confirmed that the prevalence of severe hypoglycemia remains a major obstacle to attempts to prevent diabetic complications with intensive insulin therapy. Until glucose counterregulation is more fully understood and methods for preventing hypoglycemia developed, patients with recurrent hypoglycemia unawareness or a history of hypoglycemia-related accidents should probably not be treated with intensive insulin therapy.