Limited Impact of Vigorous Exercise on Defenses Against Hypoglycemia: Relevance to Hypoglycemia-Associated Autonomic Failure

Thirteen healthy young adults gave their informed consent to participate in this study, which was approved by the Washington University Human Studies Committee and conducted at the Washington University General Clinical Research Center (GCRC). Four were women; nine were men. Their mean (±SD) age was 23.3 ± 2.4 years. Their mean BMI was 23.3 ± 3.8 kg/m².

Subjects were studied on three separate occasions in random sequence: 1) cycle exercise at 70% peak oxygen consumption (Vo_2peak) from 0830 to 0930 h and from 1200 to 1300 h on day 1 and hyperinsulinemic (2.0 mU · kg⁻¹ · min⁻¹; 12 pmol · kg⁻¹ · min⁻¹) stepped hypoglycemic clamps (hourly steps at ∼85, 75, 65, 55, and 45 mg/dl; 4.7, 4.2, 3.6, 3.1, and 2.5 mmol/l) (24) the morning of day 2, 2) rest on day 1 and identical hyperinsulinemic stepped hypoglycemic clamps on day 2- and 3) hyperinsulinemic euglycemic (∼90 mg/dl, 5.0 mmol/l) clamps.

Before entry into the study, all potential subjects were screened to ensure that they met the inclusion criteria—good health on the basis of a medical history and physical examination, normal hematocrits, fasting plasma glucose concentrations, and electrocardiograms—and Vo_2peak was determined as described previously (25). On the exercise and rest days (day 1), the subjects reported to the GCRC at ∼0730 h, a line was inserted into an antecubital vein (for sampling), and they either exercised on a cycle ergometer at ∼70% Vo_2peak from 0830 to 0930 h and again from 1200 to 1300 h or rested in the sitting position. Blood samples were obtained at 30-min intervals from 0830 through 1030 h and from 1200 through 1400 h. A snack was provided at 1100 h. On the next day (day 2), the subjects reported to the GCRC after an overnight fast at ∼0630 h. An intravenous line (for insulin and glucose infusions) and a line in a hand vein (with that hand kept in an ∼60°C plexiglas box for arterialized venous blood sampling) were inserted, and electrocardiogram leads and a vital signs monitor (Propaq Encore; Protocol Systems, Beverton, OR) were attached. The subjects remained supine throughout the study. After 30 min of supine rest and starting at ∼0730 h, regular insulin was infused in a dose of 2.0 mU · kg⁻¹ · min⁻¹ (12.0 pmol · kg⁻¹ · min⁻¹) from 0 through 300 min. Glucose (20%) was infused at variable rates, based on plasma glucose measurements with a glucose oxidase method (Yellow Springs Analyzer 2; Yellow Springs Instruments, Yellow Springs, OH) every 5 min, to maintain plasma glucose concentrations at target levels of 85, 75, 65, 55, and 45 mg/dl (4.7, 4.2, 3.6, 3.1, and 2.5 mmol/l) in hourly steps (24). The euglycemic clamps, on a separate occasion, were identical except that plasma glucose concentrations were held at ∼90 mg/dl (5.0 mmol/l). Arterialized venous samples for analytes (listed below) other than glucose and symptom scores were obtained at 30-min intervals throughout. Heart rates and blood pressures were recorded at 15-min intervals; the electrocardiogram was monitored throughout.

Plasma insulin (26), C-peptide (26), glucagon (27), pancreatic polypeptide (28), growth hormone (29), and cortisol (30) were measured with radioimmunoassays. Plasma epinephrine and norepinephrine were measured with a single isotope derivative (radioenzymatic) method (31). Serum nonesterified fatty acids (32) and blood β-hydroxybutyrate (33), lactate (34), and alanine (35) were measured with enzymatic methods. Symptoms of hypoglycemia were quantified by asking the subjects to score (0 [none] to 6 [severe]) each of 12 symptoms: six neurogenic symptoms (adrenergic: heart pounding, shaky/tremulous, and nervous/anxious; cholinergic: sweaty, hungry, and tingling) and six neuroglycopenic symptoms (difficulty thinking/confused, tired/drowsy, weak, warm, faint, and dizzy) based on our published data (36).

Data in this manuscript are reported as the means ± 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.

Cycle exercise targeted at 70% Vo_2peak (37 ± 6 [SD] ml · kg⁻¹ · min⁻¹) raised oxygen consumption to 69 ± 1% Vo_2peak from 0830 to 0930 h and to 67 ± 2% Vo_2peak from 1200 to 1300 h on the exercise day 1. As shown in Fig. 1, it raised plasma cortisol concentrations to 16.5 ± 2.0 μg/dl (455 ± 55 nmol/l) compared with 11.2 ± 1.5 μg/dl (310 ± 40 nmol/l) at the same time on the rest day 1 and to 16.6 ± 1.6 μg/dl (460 ± 45 nmol/l) compared with 8.7 ± 0.7 μg/dl (240 ± 20 nmol/l) at the same time on the rest day 1 at 0930 h and 1300 h, respectively.

Target plasma glucose concentrations were achieved during the hyperinsulinemic stepped hypoglycemic and euglycemic clamps (Fig. 2). Plasma insulin concentrations were comparable (P = 0.2252), ∼100 μU/ml (600 pmol/l), during all three hyperinsulinemic clamps (Fig. 3). Plasma C-peptide concentrations declined, from 1.7 ± 0.3 ng/ml (0.6 ± 0.1 nmol/l) to 1.1 ± 0.1 ng/ml (0.4 ± 0.0 nmol/l), during hyperinsulinemic euglycemia and to a greater extent (P < 0.0001) during hyperinsulinemic hypoglycemia, to 0.3 ± 0.1 ng/ml (0.1 ± 0.0 nmol/l) and 0.1 ± 0.0 ng/ml (<0.1 ± 0.0 nmol/l), on the days after exercise and after rest, respectively (Fig. 3).

The plasma epinephrine response (P < 0.0001) to hyperinsulinemic hypoglycemia was reduced slightly but significantly (P = 0.0113) on the day after exercise compared with the day after rest (Fig. 4). The final values (300 min, glucose ∼45 mg/dl) were 550 ± 94 pg/ml (3,000 ± 510 pmol/l) on the day after exercise and 765 ± 106 pg/ml (4,180 ± 580 pmol/l) on the day after rest compared with a final value (300 min, glucose ∼90 mg/dl) of 30 ± 6 pg/ml (160 ± 30 pmol/l) during hyperinsulinemic euglycemia.

The plasma norepinephrine response (P = 0.0003) to hyperinsulinemic hypoglycemia was unaltered (P = 0.6270) by exercise on the previous day (Fig. 5). The final values (300 min, glucose ∼45 mg/dl) were 296 ± 28 pg/ml (1.75 ± 0.17 nmol/l) on the day after exercise and 315 ± 26 pg/ml (1.86 ± 0.15 nmol/l) on the day after rest compared with a final value (300 min, glucose ∼90 mg/dl) of 205 ± 28 pg/ml (1.21 ± 0.17 nmol/l) during hyperinsulinemic euglycemia.

The neurogenic (Fig. 6) and neuroglycopenic (Fig. 7) symptom responses (P = 0.0009 and <0.0001, respectively) to hyperinsulinemic hypoglycemia were also unaltered (P = 0.6470 and 0.6624, respectively) by exercise on the previous day. The final (300 min, glucose ∼45 mg/dl) neurogenic symptom scores were 8.6 ± 2.1 on the day after exercise and 8.3 ± 1.6 on the day after rest compared with a final score (300 min, glucose ∼90 mg/dl) of 2.9 ± 0.6 during hyperinsulinemic euglycemia. The final (300 min, glucose ∼45 mg/dl) neuroglycopenic symptom scores were 6.5 ± 1.8 on the day after exercise and 6.2 ± 1.9 on the day after rest compared with a final score (300 min, glucose ∼90 mg/dl) of 2.2 ± 1.0 during hyperinsulinemic euglycemia.

The plasma glucagon response (P < 0.0001) to hyperinsulinemic hypoglycemia was not reduced by exercise on the previous day (Fig. 8); indeed, the glucagon levels were slightly higher (P = 0.0436) on the day after exercise. The final values (300 min, glucose ∼45 mg/dl) were 88 ± 7 pg/ml (25 ± 2 pmol/l) on the day after exercise and 80 ± 8 pg/ml (23 ± 2 pmol/l) on the day after rest compared with a final value (300 min, glucose ∼90 mg/dl) of 37 ± 2 pg/ml (11 ± 1 pmol/l) during hyperinsulinemic euglycemia.

The glucose infusion rates required to maintain the plasma glucose steps during hyperinsulinemic hypoglycemia were slightly but significantly (P = 0.0108) higher on the day after exercise compared with the day after rest (Fig. 9). The final values (300 min, glucose ∼45 mg/dl) were 4.7 ± 0.6 mg · kg⁻¹ · min⁻¹ (26 ± 3 μmol · kg⁻¹ · min⁻¹) on the day after exercise and 2.4 ± 0.6 mg · kg⁻¹ · min⁻¹ (13 ± 3 μmol · kg⁻¹ · min⁻¹) on the day after rest compared with a final value (300 min, glucose ∼90 mg/dl) of 13.0 ± 0.8 mg · kg⁻¹ · min⁻¹ (72 ± 4 μmol · kg⁻¹ · min⁻¹) during hyperinsulinemic euglycemia.

The plasma growth hormone response (P < 0.0001) but not the plasma pancreatic polypeptide or cortisol response both (P < 0.0001) to hypoglycemia was reduced significantly on the day after exercise compared with the day after rest (Table 1). The P values were 0.0090 for growth hormone, 0.0629 for pancreatic polypeptide, and 0.1275 for cortisol. There were no sex differences. However, only four women were studied.

Serum nonesterified fatty acid (P = 0.3900) and blood β-hydroxybutyrate (P = 0.8465) concentrations were suppressed comparably under all three study conditions (Table 2). Increments in blood lactate levels were similar (0.3064) during hypoglycemia on the days after exercise and rest (Table 2). Similarly, there was no difference in blood alanine levels (P = 0.1921).

Heart rates (P = 0.0788) and systolic blood pressures (P = 0.3304) were similar under all three study conditions (Table 3). Diastolic blood pressures declined (P = 0.0074) during the hypoglycemic clamps, but there was no difference after exercise compared with after rest (P = 0.6678).

These data demonstrate that two bouts of vigorous cycle exercise—69 ± 1% and 67 ± 2% of peak oxygen consumption from 0830 to 0930 h and from 1200 to 1300 h, respectively—raised plasma cortisol concentrations during exercise and reduced the plasma epinephrine and growth hormone responses to hyperinsulinemic stepped hypoglycemia on the next day slightly but significantly. Plasma norepinephrine, neurogenic and neuroglycopenic symptom and plasma pancreatic polypeptide, glucagon, and cortisol responses to hypoglycemia were not reduced by exercise on the previous day.

These findings differ quantitatively and in many respects qualitatively, from those of Galassetti et al. (23), who assessed the impact of two somewhat longer bouts of moderate exercise (∼50% Vo_2peak for 90 min) on responses to hyperinsulinemic hypoglycemia (54 mg/dl, 3.0 mmol/l) on the next day. First, in the present study, the impact of previous exercise on the epinephrine and growth hormone responses to subsequent hypoglycemia was small. Clearly, these responses were not eliminated; the data suggest a small shift of the glycemic thresholds for these responses to slightly lower plasma glucose concentrations. Second, in contrast to the findings of Galassetti et al. (23), the present data show no effect of previous vigorous exercise on the plasma norepinephrine, pancreatic polypeptide, and glucagon responses to subsequent hypoglycemia. The reasons for these differences are not entirely clear. The day 1 exercise patterns differed. Ours was more vigorous, theirs was somewhat longer. Nonetheless, on balance it would seem that the exercise stimuli were similar. Our final hypoglycemic step was shorter than their hypoglycemic clamp, but our two steps, at 55 and 45 mg/dl over 2 h, would have produced a comparable if not more potent hypoglycemic stimulus. Our subjects were somewhat more fit. Their exercise protocol resulted in higher absolute plasma cortisol concentrations. We suspect that this is an important difference. It may be relevant that the control (rest) data in the present study were obtained in the same subjects studied concurrently in random sequence with the exercise study and under identical conditions aside from rest or exercise on day 1.

Despite these seemingly substantive quantitative differences, there is some qualitative agreement with respect to the impact of previous exercise on the autonomic responses (as well as on the symptomatic responses) to hypoglycemia. Both studies indicate that the adrenomedullary (epinephrine) response to hypoglycemia is reduced to some extent after exercise on the previous day. Although the present data do not indicate that the sympathetic neural (norepinephrine) response is reduced, Galassetti et al. (23) found the muscle sympathetic nerve activity response (as well as the plasma norepinephrine response) to be reduced. The microneurographic measurement of muscle sympathetic nerve activity is probably a more sensitive measure of sympathetic neural activation than the plasma norepinephrine concentration (37).

Thus, the data are qualitatively consistent with the hypothesis of Davis et al. (20,21) that cortisol at least in part mediates the reduced autonomic component of HAAF (10–14). However, the limited impact of previous exercise and the resulting transient increase in plasma cortisol on the key glucose counterregulatory hormone responses to subsequent hypoglycemia—no effect on the glucagon response and a small reduction of the epinephrine response in the present data—suggest that an additional factor or factors may well be involved in the pathogenesis of the reduced autonomic component of HAAF.

Although the biological impact of the observed small reduction in the epinephrine response to hypoglycemia after exercise on the previous day is open to question, significantly higher glucose infusion rates were required to maintain the hypoglycemic clamps on the day after exercise. That finding indicates increased responsiveness to insulin on the day after exercise. The extent to which that was the result of the reduced epinephrine response or some other mechanism is unknown.

In contrast to the impact of previous exercise on the epinephrine response to hypoglycemia, exercise had no effect on the symptomatic responses to hypoglycemia on the next day. Importantly, neurogenic (autonomic) symptom scores—largely the result of the perception of physiologic changes caused by the autonomic (adrenomedullary and sympathetic neural) discharge triggered by hypoglycemia (36)—were no different during hypoglycemia on the day after exercise from those during hypoglycemia on the day after rest. Galassetti et al. (23) also found no effect of previous exercise on the symptomatic responses to hypoglycemia. Thus, it seems that the reduced neurogenic symptom component of HAAF (20–22) is not mediated by cortisol.

In summary, the present data provide some additional support for the hypothesis that the cortisol response to hypoglycemia mediates in part the reduced autonomic response to subsequent hypoglycemia (20–22), one key component of the clinical concept of HAAF in type 1 diabetes (10–16). However, the small effect of vigorous exercise-induced cortisol release on the epinephrine response—with no significant effect on the norepinephrine or pancreatic polypeptide responses—to subsequent hypoglycemia observed suggests that an additional factor or factors may well be involved. Furthermore, the present and previous (23) data do not support the hypothesis that the cortisol response to hypoglycemia mediates the reduced neurogenic symptom response to subsequent hypoglycemia, another key component of the concept of HAAF in type 1 diabetes (10–14).

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

The authors acknowledge the technical assistance of Suresh Shah, Krishan Jethi, Shobna Mehta, Joy Brothers, Carolyn Fritsche, Zina Lubovich, Michael Morris, Sharon O’Neill, Lennis Lich, Suzanne Fritsch, and John Hood, Jr.; 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.