Hypoglycemia-Associated Autonomic Failure in Advanced Type 2 Diabetes

A total of 13 patients with type 2 diabetes—7 treated with oral hypoglycemic agents (OHA R_X) and 6 treated with insulin—and 15 nondiabetic control subjects gave their written informed consent to participate in this study, which was approved by the Washington University Studies Committee and conducted at the Washington University General Clinical Research Center. The characteristics of the patients and the control subjects are listed in Table 1.

Exclusion criteria included the presence of circulating GAD antibodies, anemia, untreated hypertension, untreated proliferative retinopathy, renal insufficiency, autonomic neuropathy and clinically overt coronary, cerebral or peripheral macrovascular disease, and any additional contraindication to experimental hypoglycemia, such as known central nervous system disease including a history of seizures. Patients who were treated with OHA R_X were using a sulfonylurea, metformin or a thiazolidinedione, or a combination of these. Patients who were treated with insulin were selected for a relatively long-standing requirement for therapy with insulin. The intent was not to contrast impacts of the types of therapy but rather to use the requirement for long-term insulin therapy as an index of advanced type 2 diabetes. The effectiveness of that selection process is supported by the lower fasting plasma C-peptide concentrations in the patients who were treated with insulin (P = 0.0051 vs. nondiabetic subjects and 0.0172 vs. OHA R_X type 2 diabetic patients, Table 1).

The participants were studied, after overnight fasts, on two consecutive days. Hyperinsulinemic (2.0 mU · kg⁻¹ · min⁻¹, 12.0 pmol · kg⁻¹ · min⁻¹) stepped hypoglycemic clamps (29) (85, 75, 65, 55 and 45 mg/dl, 4.7, 4.2, 3.6, 3.3, and 2.5 mmol/l, in hourly steps) were performed on the morning of day 1 and of day 2 with an additional 2-h period of hypoglycemia (50 mg/dl, 2.8 mmol/l) on the afternoon of day 1. Arterialized venous samples, with the participants in the supine position throughout, were drawn from a line inserted in a hand vein (with that hand kept in an ∼60°C plexiglas box) at −30, −15, and 0 min and at 30-min intervals through 300 min. Insulin and glucose (20%) were infused through a line in an antecubital vein with the glucose infusion rate varied, based on plasma glucose measurements every 5 min, to clamp glucose levels at the target levels. Blood pressure and heart rate were also recorded (Propaq, Encore, Protocol Systems, Beverton, OR) at 30-min intervals; the electrocardiogram was monitored throughout.

Plasma glucose was measured with a glucose oxidase method (Glucose Analyzer 2; Beckman, Brea, CA). Plasma insulin (30), C-peptide (30), and glucagon (31) were measured with radioimmunoassays, plasma epinephrine, and norepinephrine with a single isotope derivative (radioenzymatic) method (32). Symptoms of hypoglycemia were quantified by asking the participants to score (0, none to 6, severe) each of 12 symptoms: 6 neurogenic symptoms (adrenergic: heart pounding, shaky/tremulous, and nervous/anxious; cholinergic: sweaty, hungry, and tingling) and 6 neuroglycopenic symptoms (difficulty thinking/confused, tired/drowsy, weak, warm, faint and dizzy) based on our published data (33).

Data are expressed as the mean ± SE except where the SD is specified. Data were analyzed by general linear model repeated measures ANOVA. P < 0.05 was considered to indicate statistical significance.

Target plasma glucose concentrations were approximated during the hyperinsulinemic stepped hypoglycemic clamps in all three groups of participants on days 1 and 2 (Fig. 1, Table 2). Plasma insulin concentrations were higher in both groups of patients with type 2 diabetes than in the nondiabetic subjects, but there were no significant differences between the day 1 and day 2 insulin levels during the hyperinsulinemic clamps in any of the three groups of participants (Table 3).

Compared with that of nondiabetic control subjects, the plasma glucagon response to hypoglycemia was reduced (P = 0.0252) in insulin R_x type 2 diabetic patients but not in OHA R_x type 2 diabetic patients (Fig. 2, Table 4). Indeed, the glucagon response was virtually absent in the type 2 diabetes insulin R_X group.

The plasma epinephrine response to hypoglycemia was increased in the type 2 diabetes OHA R_X group (P = 0.0082 versus nondiabetic), i.e., it occurred at higher plasma glucose concentrations (Fig. 3, Table 5). Although the epinephrine response seemed to be reduced at the lowest plasma glucose concentration in the type 2 diabetes insulin R_X group (Fig. 3), that seeming difference was not statistically significant. The plasma norepinephrine response to hypoglycemia was also increased in the type 2 diabetes OHA R_X group (P = 0.0082 versus nondiabetic; Fig. 4, Table 6).

The neurogenic (autonomic) symptom response to hypoglycemia was increased in the type 2 diabetes OHA R_X group (P = 0.0023 versus nondiabetic; Fig. 5, Table 7). The neuroglycopenic symptom response to hypoglycemia was also increased in the type 2 diabetes OHA R_X group (P = 0.0182 versus nondiabetic; Table 8).

Hypoglycemia on day 1 was associated with reduced plasma glucagon (P = 0.0249; Table 4), plasma epinephrine (P = 0.0001; Table 5), plasma norepinephrine (P = 0.0106; Table 6), and neuroglycopenic symptom (P = 0.0467; Table 8) responses to hypoglycemia on day 2 in nondiabetic subjects. The neurogenic symptom response tended to be reduced (P = 0.1488; Table 7).

In the patients with type 2 diabetes as a group (n = 13), hypoglycemia on day 1 was associated with reduced plasma glucagon (P = 0.0015), plasma epinephrine (P = 0.0002), plasma norepinephrine (P = 0.0138), neurogenic symptom (P = 0.0149), and neuroglycopenic symptom (P = 0.0015) responses to hypoglycemia on day 2 (group data not shown).

Hypoglycemia on day 1 was associated with reduced plasma glucagon (P = 0.0036; Fig. 6, Table 4), plasma epinephrine (P = 0.0022; Fig. 7, Table 5), plasma norepinephrine (P = 0.0441; Fig. 8, Table 6), neurogenic symptom (P = 0.0306; Fig. 9, Table 7), and neuroglycopenic symptom (P = 0.0097; Table 8) responses on day 2 in the type 2 diabetes OHA R_X group.

Hypoglycemia on day 1 was associated with apparently but not significantly reduced plasma glucagon (P = 0.0821; Table 4), plasma epinephrine (P = 0.0740; Table 5), and plasma norepinephrine (P = 0.1434; Table 6) responses to hypoglycemia on day 2 in the type 2 diabetes insulin R_X group. Neurogenic symptom (P = 0.0202; Table 7) and neuroglycopenic symptom (P = 0.0082; Table 8) responses to hypoglycemia were reduced significantly.

There were no significant differences in heart rate or in systolic or diastolic blood pressure among the three groups (data not shown).

These data support our hypotheses that the glucagon response to hypoglycemia is reduced in patients who are approaching the insulin-deficient end of the spectrum of type 2 diabetes and that recent antecedent hypoglycemia shifts glycemic thresholds for autonomic (including adrenomedullary epinephrine) and symptomatic responses to subsequent hypoglycemia to lower plasma glucose concentrations in type 2 diabetes. They also indicate that the glycemic thresholds for these responses are shifted to higher plasma glucose concentrations at baseline (day 1) in patients with poorly controlled type 2 diabetes.

An absent glucagon response to falling plasma glucose concentrations (2–4) is a key pathophysiological feature of the clinical syndrome of defective glucose counterregulation (1,8,9) and the concept of hypoglycemia-associated autonomic failure in type 1 diabetes (1,7,10,11). Because the absent glucagon response is linked tightly with the loss of endogenous insulin secretion in type 1 diabetes (4), we reasoned that the glucagon response might be reduced in patients who are approaching the insulin-deficient end of the spectrum of type 2 diabetes. Accordingly, we selected patients who required insulin therapy for management of their type 2 diabetes for an average of 5 years, a clinical surrogate for insulin deficiency that was supported by the finding of significantly reduced plasma C-peptide levels. Indeed, the mean fasting plasma C-peptide concentration of 1.1 ng/ml in these patients was lower than the optimal cutoff value of 1.3 ng/ml or less found by Berger et al. (34) to distinguish type 1 diabetes from typical type 2 diabetes. We also required the absence of circulating GAD antibodies to exclude patients with late-onset type 1 diabetes (35,36). Thus, our selection process effectively identified patients with insulin-deficient type 2 diabetes. In these patients with advanced type 2 diabetes, the glucagon response to hypoglycemia was virtually absent, just as it is in type 1 diabetes (2–4). This is a statistically robust finding because it was documented with a rather small sample of appropriately identified patients whose glucagon responses were compared with those of age-, gender-, and BMI-matched nondiabetic control subjects.

Plasma insulin concentrations were higher at baseline and during the hyperinsulinemic stepped hypoglycemic clamps in both groups of patients with type 2 diabetes, although there were no significant differences between day 1 and day 2 insulin levels in any of the three groups. These supraphysiological insulin levels permitted accurate matching of plasma glucose concentrations at euglycemic and progressively hypoglycemic (to 45 mg/dl) levels. Such matching of the stimulus is critical to the interpretation of the neuroendocrine and symptomatic responses to hypoglycemia. However, hyperinsulinemia per se cannot explain the reduced glucagon response to hypoglycemia in the insulin-deficient (type 2 diabetic insulin R_x) patients because there was a normal glucagon response to hypoglycemia in the insulin-sufficient (type 2 diabetes OHA R_x) patients despite comparable hyperinsulinemia. We also considered the possibility that unrecognized recent antecedent iatrogenic hypoglycemia might have caused the reduced glucagon response to hypoglycemia in the (insulin-treated) insulin-deficient patients. Because the patients were studied as General Clinical Research Center inpatients, with frequent plasma glucose measurements overnight before the glucose clamps, we know that hypoglycemia did not occur the night before the study. We cannot be absolutely certain that it did not occur before then. However, the fact that the epinephrine and symptomatic responses to hypoglycemia—sensitive measures of the impact of recent antecedent hypoglycemia (7,10,11,37,38)—were not reduced significantly on day 1 provides strong evidence that recent antecedent hypoglycemia was not the cause of the reduced glucagon response.

An attenuated epinephrine response to falling plasma glucose concentrations—in the setting of an absent glucagon response—causes the clinical syndrome of defective glucose counterregulation in type 1 diabetes (1,8,9). Attenuated autonomic (including adrenomedullary epinephrine) and symptomatic responses to falling plasma glucose concentrations cause the clinical syndrome of hypoglycemia unawareness in type 1 diabetes (1). These are the central features of the concept of hypoglycemia-associated autonomic failure in type 1 diabetes (1,7,10,11). The autonomic—adrenomedullary (epinephrine), sympathetic neural (norepinephrine), and neurogenic (autonomic) symptom—responses to hypoglycemia are not eliminated; rather, the glycemic thresholds for these responses are shifted to lower plasma glucose concentrations. In other words, these responses can be elicited, but lower plasma glucose concentrations are required to elicit them. These shifts in glycemic thresholds are largely the result of recent antecedent iatrogenic hypoglycemia (1,7,10,11). Because this phenomenon is demonstrable in healthy individuals (1,37,38) and in those with type 1 diabetes (1,7,10,11), it is reasonable to anticipate that it would be demonstrable in those with type 2 diabetes. The present data document that expectation. Plasma epinephrine and norepinephrine responses and neurogenic (as well as neuroglycopenic) symptom responses to an identical hypoglycemic stimulus were reduced significantly on the day after two episodes of hypoglycemia in patients with type 2 diabetes just as they are in type 1 diabetes (7,10,11). Notably, these responses were not eliminated. Rather, lower plasma glucose concentrations were required to elicit them, i.e., the glycemic thresholds for the adrenomedullary and sympathetic neural responses and for the resultant neurogenic symptom responses were shifted to lower plasma glucose concentrations in patients with type 2 diabetes just as they are in type 1 diabetes (7,10,11).

Albeit demonstrable in the patients with type 2 diabetes as a whole, the effect of recent antecedent hypoglycemia to shift the glycemic thresholds for autonomic and symptomatic responses to hypoglycemia to lower plasma glucose concentrations was most clearly demonstrated in the patients with relatively poorly controlled type 2 diabetes (mean HbA_1c of 8.6%) treated with OHA. In those patients, the plasma epinephrine and norepinephrine and neurogenic (and neuroglycopenic) symptom responses to hypoglycemia were enhanced compared with those of nondiabetic control subjects at baseline (day 1). These responses occurred at higher-than-normal plasma glucose concentrations. Spyer et al. (39) reported a similar pattern. Thus, glycemic thresholds for these responses are shifted to higher plasma glucose concentrations in patients with poorly controlled type 2 diabetes, just as they are in poorly controlled type 1 diabetes (40,41).

We did not study glucose recovery from hypoglycemia in type 2 diabetes. However, because the contribution of absent glucagon and attenuated epinephrine responses to falling plasma glucose concentrations has been shown in prospective studies to increase the risk of severe iatrogenic hypoglycemia 25-fold or more during aggressive therapy of type 1 diabetes (8,9), it is reasonable to suggest that patients with advanced type 2 diabetes, shown here to exhibit this pattern of counterregulatory abnormalities, are also at increased risk. If so, one could question the extent to which the higher frequency of iatrogenic hypoglycemia in insulin-treated type 2 diabetes (20,21) is the result of the greater glucose-lowering potency of insulin (given in sufficient doses) or of the impact of compromised glucose counterregulation in advanced type 2 diabetes demonstrated in the present data or both.

Because the glucagon response to falling plasma glucose concentrations is virtually absent and the glycemic thresholds for autonomic (including adrenomedullary epinephrine) and symptomatic responses to hypoglycemia are shifted to lower plasma glucose concentrations by recent antecedent hypoglycemia, patients with advanced type 2 diabetes, like those with type 1 diabetes, are at risk for defective glucose counterregulation and hypoglycemia unawareness, the components of hypoglycemia-associated autonomic failure and the resultant vicious cycle of recurrent iatrogenic hypoglycemia. This may explain why iatrogenic hypoglycemia becomes limiting to glycemic control as patients approach the insulin-deficient end of the spectrum of type 2 diabetes.

This work was supported, in part, by U.S. Public Health Service Grants R01-DK27085, M01-RR00036, P60-DK20579, and T32-DK07120 and by a fellowship award from the American Diabetes Association.

We acknowledge the technical assistance of Suresh Shah, Krishan Jethi, Mary Hamilton, Joy Brothers, Carolyn Fritsche, Zina Lubovich, Michael Morris, Sharon O’Neill, and Shobhna Mehta; the assistance of the nursing staff of the Washington University Clinical Research Center; and the assistance of Karen Muehlhauser in the preparation of this manuscript.