Sleep-Related -Associated Autonomic Failure in Type 1 Reduced Awakening From Sleep During Hypoglycemia Salomon Banarer and Philip E. Cryer

Given that iatrogenic hypoglycemia often occurs during the night in people with , we tested the hypothesis that physiological, and the resulting behav- atrogenic hypoglycemia is the limiting factor in the ioral, defenses against developing hypoglycemia—al- glycemic management of diabetes, both conceptu- ready compromised by absent and attenuated ally and in practice (1). It causes recurrent and epinephrine and neurogenic symptom responses—are sometimes permanent physical morbidity, recurrent further compromised during sleep in type 1 diabetes. To I or persistent psychosocial morbidity, and occasionally do so, we studied eight adult patients with uncompli- cated type 1 diabetes and eight matched nondiabetic death, and it precludes true long-term glycemic control in control subjects with hyperinsulinemic stepped hypo- most patients with type 1 diabetes (2) and many with type glycemic clamps ( steps of ϳ85, 75, 65, 55, and 45 2 diabetes (3). Thus long-term complications of diabetes mg/dl) in the morning (0730–1230) while awake and at can occur despite aggressive attempts to achieve glycemic night (2100–0200) while awake throughout and while control (2,3). Iatrogenic hypoglycemia is the result of the asleep from 0000 to 0200 in random sequence. Plasma interplay between relative or absolute therapeutic .(excess and compromised glucose counterregulation (1 ؍ perhaps norepinephrine (P ,(0.0010 ؍ epinephrine (P ؍ 0.0838), and pancreatic polypeptide (P 0.0034) re- The concept of hypoglycemia-associated autonomic fail- sponses to hypoglycemia were reduced during sleep in ure (HAAF) in type 1 diabetes (1,4,6) and advanced type 2 diabetic subjects (the final awake versus asleep values were 240 ؎ 86 and 85 ؎ 47, 205 ؎ 24 and 148 ؎ 17, and diabetes (1,5,6) posits that recent antecedent iatrogenic and 118 ؎ 31 pg/ml, respectively), but not in hypoglycemia causes defective glucose counterregulation 45 ؎ 197 the control subjects. The diabetic subjects exhibited (by reducing the epinephrine response to subsequent markedly reduced awakening from sleep during hypo- hypoglycemia in the setting of an absent glucagon re- glycemia. Sleep efficiency (percent time asleep) was sponse) and hypoglycemia unawareness (by reducing the -autonomic [sympathetic neural and adrenomedullary] re ؍ in the diabetic subjects, but only 26 ؎ 8% (P 18% ؎ 77 0.0109) in the control subjects late in the 45-mg/dl sponse and therefore the neurogenic symptom responses hypoglycemic steps. We conclude that autonomic re- to subsequent hypoglycemia) and thus a vicious cycle of sponses to hypoglycemia are reduced during sleep in recurrent iatrogenic hypoglycemia. There is considerable type 1 diabetes, and that, probably because of their support for the concept of HAAF and its clinical impact, reduced sympathoadrenal responses, patients with type 1 diabetes are substantially less likely to be awakened including the finding that as little as 2–3 weeks of scrupulous by hypoglycemia. Thus both physiological and behav- avoidance of iatrogenic hypoglycemia reverses hypoglyce- ioral defenses are further compromised during sleep. mia unawareness and improves the reduced epinephrine This sleep-related hypoglycemia-associated autonomic component of defective glucose counterregulation in most failure, in the context of imperfect insulin replacement, affected patients (rev. in 1). The mediator(s) and mecha- likely explains the high frequency of nocturnal hypogly- nism(s) of HAAF are under active investigation (1,6). cemia in type 1 diabetes. Diabetes 52:1195–1203, 2003 Iatrogenic hypoglycemia, including severe hypoglyce- mia, often occurs during sleep (2,7,8), but the physiology of glucose counterregulation and its pathophysiology in type 1 diabetes during the night, and specifically during sleep, have not been extensively studied. Bendtson et al. (9), studying adults with type 1 diabetes, reported en- hanced plasma epinephrine, norepinephrine, and cortisol responses to hypoglycemia induced at night compared with that induced in the morning. However, plasma glu- From the Division of , Metabolism, and Lipid Research, General Clinical Research Center and Diabetes Research and Training Center, Wash- cose concentrations were not clamped and remained at a ington University School of Medicine, St. Louis, Missouri. lower level longer in the nocturnal study. Furthermore, the Address correspondence and reprint requests to Philip E. Cryer, M.D., Campus Box 8127, Washington University School of Medicine, 660 S. Euclid effects of sleep per se were not studied, and the extent to Ave., St. Louis, MO 63110. E-mail: [email protected]. which the patients were awake or asleep was not reported. Received for publication 13 November 2002 and accepted in revised form 10 In their study of adolescents with type 1 diabetes, Jones et February 2003. HAAF, hypoglycemia-associated autonomic failure. al. (10) found reduced plasma epinephrine, norepineph- © 2003 by the American Diabetes Association. rine, and cortisol responses to brief nocturnal hypoglyce-

DIABETES, VOL. 52, MAY 2003 1195 SLEEP AND HYPOGLYCEMIA

TABLE 1 Nondiabetic subjects were admitted 1 h before each study. Studies in any one Characteristics of nondiabetic subjects and patients with type 1 given subject were separated by at least 2 weeks. diabetes Experimental design. Intravenous lines were inserted into an antecubital vein (for insulin and glucose infusions) and into a hand vein, with that hand Nondiabetic Diabetic kept in a 55°C Plexiglas box (for arterialized venous blood sampling) at Ϫ60 to Ϫ30 min. The subjects remained supine from Ϫ30 through 300 min. Sex (female/male) 3/5 3/5 Hyperinsulinemic (2.0 mU ⅐ kgϪ1 ⅐ minϪ1) stepped hypoglycemic clamps Age (years) 27.4 Ϯ 7.1 27.0 Ϯ 7.4 (hourly steps at 85, 75, 65, 55, and 45 mg/dl) (11) were performed on three BMI (kg/m2) 26.0 Ϯ 1.4 26.4 Ϯ 3.7 occasions in random, computer-generated sequence in both groups. These Ϯ Ϯ were performed in the morning (0730–1230) and twice during the night HbA1c (%) 5.1 0.4 8.3 1.2 Ϯ (2100–0200), once with the subject awake and once with the subject allowed Duration of diabetes (years) — 10.6 7.4 ⅐ Ϫ1 ⅐ Ϫ1 Ϯ to sleep starting at 1130 h. Hyperinsulinemic (2.0 mU kg min ), Insulin dosage (units/day) — 33 16 euglycemic (85 mg/dl) clamps (11) were performed in the morning on a fourth Data are n or means Ϯ SD. occasion in the nondiabetic subjects. These clamps were accomplished by variable intravenous infusions of 20% dextrose based on arterialized venous plasma glucose measurements every 5 min at bedside (Yellow Springs mia when the patients were asleep compared with when Analyzer 2; Yellow Springs Instruments, Yellow Springs, OH). Arterialized they were awake. The growth hormone response was not venous blood samples for the analytes listed below were drawn at Ϫ15 and 0 altered during sleep; glucagon and pancreatic polypeptide min and then every 30 min through 300 min. Blood pressures and heart rates (Propaq Encore; Protocol Systems, Beverton, OR) were recorded at Ϫ15 and responses were not reported. Age-matched nondiabetic 0 min and every 30 min through 300 min. The electrocardiogram was subjects were not studied while awake at night, but their monitored throughout. Symptoms of hypoglycemia were quantified, also at plasma epinephrine, norepinephrine, and cortisol re- 30-min intervals, by asking the subjects to score on a scale of zero (none) to sponses to hypoglycemia were reduced while they were six (severe) each of 12 symptoms: 6 neurogenic symptoms (adrenergic: heart pounding, shaky/tremulous, and nervous/anxious; cholinergic: sweaty, hun- asleep at night compared with the responses in the day- gry, and tingling) and 6 neuroglycopenic symptoms (difficulty thinking/ time while they were awake. Growth hormone and gluca- confused, tired/drowsy, weak, warm, faint, and dizzy) based on our published gon responses were not reduced during sleep; the data (12). Symptoms were not, of course, assessed during sleep. pancreatic polypeptide response was not reported. Polysomnographic recordings for determination of sleep stages were made To date, we have attributed HAAF in diabetes entirely to using 14 electrodes (including left and right electrooculograms, electromyo- grams, and four scalp electroencephalograms). Staging was hand scored by a recent antecedent iatrogenic hypoglycemia (1,4–6). How- single technician using the Rechtschaffen and Kales (13) standardized scoring ever, given the findings of Jones et al. (10) in adolescents for each 30-s epoch. Recordings were performed from 0000 through 0200 (i.e., with type 1 diabetes, we considered the possibility that an at nominal plasma glucose steps of 55 and 45 mg/dl) during all 16 sleep additional factor, sleep, might produce a similar phenom- studies. However, data were available for analysis only for the eight nondia- betic subjects and four of the eight diabetic patients. enon. Accordingly, we tested the hypothesis that physio- Analytical methods. Plasma glucose was measured with a glucose oxidase logical, and the resulting behavioral, defenses against method (Yellow Springs Analyzer 2). Plasma insulin (14), C-peptide (14), developing hypoglycemia (already compromised by absent glucagon (15), pancreatic polypeptide (16), growth hormone (17), and cortisol glucagon and attenuated epinephrine and neurogenic (18) were measured with radioimmunoassays. Plasma epinephrine and nor- symptom responses) are further compromised during epinephrine were measured with a single isotope derivative (radioenzymatic) method (19). Serum nonesterified fatty acid (20) and blood ␤-hydroxybutyrate sleep in adults with type 1 diabetes. To do so, we studied (21), lactate (22), and alanine (23) were measured with enzymatic methods. patients with uncomplicated type 1 diabetes and matched Statistical methods. Data in this manuscript are expressed as means Ϯ SE, nondiabetic control subjects with hyperinsulinemic except where the standard deviation is specified. Time- and condition-related stepped hypoglycemic clamps in random sequence in the data were analyzed by general linear model repeated measures ANOVA after adjustment for any differences at baseline. Sleep data were analyzed by t test. morning while awake, during the night while awake, and P Ͻ 0.05 was considered to indicate statistically significant differences. during the night while asleep. The findings are indicative of a second type of HAAF, characterized as sleep-related HAAF in diabetes. RESULTS Glucose, insulin, and C-peptide. Plasma glucose con- RESEARCH DESIGN AND METHODS centrations were clamped at target levels in both the Subjects. We studied eight patients with type 1 diabetes and eight nondia- nondiabetic and diabetic subjects (Fig. 1). Plasma insulin betic subjects matched for sex, age, and BMI; each subject gave their written concentrations were raised comparably under all condi- consent to participate. The study protocol was approved by the Washington tions in both groups (Fig. 2). Plasma C-peptide concentra- University Medical Center Human Studies Committee and conducted at the Washington University General Clinical Research Center. The characteristics tions declined during hyperinsulinemic euglycemia and to of the participants are listed in Table 1. All had normal hematocrits, serum a greater extent during hypoglycemia in the nondiabetic creatinine concentrations, and electrocardiograms, and none had a history of subjects (Fig. 2); plasma C-peptide was undetectable in the central nervous disease or cardiac arrhythmias. The diabetic patients had no diabetic subjects (Fig. 2). evidence of classical diabetic autonomic neuropathy (as evidenced by a negative medical history and physical examination including normal electro- Polysomnography. During the night asleep studies, sleep cardiographic RR variation during deep breathing and the absence of ortho- efficiency (the percent of time asleep) was significantly static hypotension) and no active . They had not higher in the diabetic than in the nondiabetic subjects experienced an episode of severe iatrogenic hypoglycemia (requiring the during the 55- and 45-mg/dl hypoglycemic steps (1200– assistance of another individual) over the 3 months before the study and had 0200) (P ϭ 0.0227), particularly during the 45-mg/dl hypo- no self-monitored blood glucose levels (measured at least three times a day) ϭ Ͻ72 mg/dl (Ͻ4.0 mmol/l) during the week before the study. (If the latter glycemic step (0100–0200) (P 0.0190) (Table 2, Fig. 3). occurred the study was postponed until that criterion was met.) Similarly, the time in sleep stages I through IV (i.e., Studies were performed after a fast of at least 10 h. Patients took their last non-REM sleep) was greater in the diabetic subjects (P ϭ prestudy dose of NPH insulin at least 12 h before each study or their last dose 0.0272 and 0.0191, respectively) (Table 2). During the final of glargine insulin at least 24 h before each study. The patients were admitted to the research center 12 h before each study. Their diabetes was managed 30 min of the 45-mg/dl hypoglycemic step, sleep efficiency with variable dosages of intravenous regular insulin to hold plasma glucose was 77 Ϯ 18% in the diabetic subjects patients and 26 Ϯ 8% concentrations in the 80–120 mg/dl range over the 10 h before each study. in the control subjects (P ϭ 0.0109) (Fig. 3).

1196 DIABETES, VOL. 52, MAY 2003 S. BANARER AND P.E. CRYER

TABLE 2 Sleep efficiency (percent of time asleep) and duration of sleep stages during the last two hypoglycemic steps (55 mg/dl [0000– 0100]) and 45 mg/dl [0100–0200]) of the night asleep studies in nondiabetic subjects and patients with type 1 diabetes Nondiabetic Diabetic n 84 0000–0100 Sleep efficiency (%) 44.4 Ϯ 13.8 66.3 Ϯ 10.0 Sleep stages (min) I 2.9 Ϯ 0.9 3.3 Ϯ 0.7 II 16.5 Ϯ 5.2 27.6 Ϯ 5.4 III 2.9 Ϯ 1.6 3.9 Ϯ 2.3 IV 4.3 Ϯ 2.3 2.5 Ϯ 2.5 REM 0.1 Ϯ 0.1 2.4 Ϯ 2.4 0100–0200 *FIG. 1. Plasma glucose concentrations (means ؎ SE) during morning Sleep efficiency (%) 34.3 Ϯ 10.2 80.0 Ϯ 10.4 hyperinsulinemic, euglycemic clamps (shaded area) and hyperinsuline- Sleep stages (min) Ϯ Ϯ ,(8 ؍ mic stepped hypoglycemic clamps in nondiabetic subjects (A; n and hyperinsulinemic stepped hypoglycemic clamps in patients with I 4.3 1.1 2.5 0.5 studied in the morning (0730–1230) while II 12.5 Ϯ 4.7 24.6 Ϯ 8.4 (8 ؍ type 1 diabetes (B; n awake (E) and during the night (2100–0200) while awake (F) and III 0.4 Ϯ 0.4 4.5 Ϯ 1.4 asleep (0000–0200; f). IV 0.0 Ϯ 0.0 11.6 Ϯ 7.0 REM 3.4 Ϯ 1.9 4.8 Ϯ 2.5 Epinephrine and norepinephrine. In both the nondia- Data are n or means Ϯ SE. *P ϭ 0.0190 vs. nondiabetic subjects; P ϭ betic and diabetic subjects, increments in plasma epineph- 0.0227 diabetic vs. nondiabetic subjects for 0000–0200. rine (Fig. 4) and norepinephrine (Fig. 5) concentrations during hypoglycemia were similar in the morning awake the 65- and 55-mg/dl glucose steps (Fig. 5). In the diabetic and night awake states. In the nondiabetic subjects, the subjects, the plasma norepinephrine response to hypogly- plasma epinephrine response to hypoglycemia at night cemia at night was not significantly reduced (P ϭ 0.0838) was not reduced significantly during sleep, although the during sleep compared with when the subjects were epinephrine levels during sleep appeared to be lower than awake, although there was no apparent response during those when the subjects were awake at the 65- and sleep (Fig. 5). 55-mg/dl glucose steps (Fig. 4). In the diabetic subjects, Neurogenic and neuroglycopenic symptoms. In both the plasma epinephrine response to hypoglycemia at night the nondiabetic and diabetic subjects, increments in neu- was reduced (P ϭ 0.0010) during sleep compared with rogenic (autonomic) (Table 3) and neuroglycopenic symp- when the subjects were awake (Fig. 4). In the nondiabetic tom scores (Table 4) during hypoglycemia were similar subjects, the plasma norepinephrine response to hypogly- during the morning awake and night awake states. Symp- cemia at night was not reduced significantly during sleep, toms were not assessed during sleep. although the norepinephrine levels during sleep appeared Glucagon. In the nondiabetic subjects, increments in to be lower than those when the subjects were awake at plasma glucagon concentrations during hypoglycemia were similar in the morning awake and night awake states (Fig. 6). The plasma glucagon response to hypoglycemia at night was not reduced during sleep (Fig. 6). In diabetic

(FIG. 2. Plasma insulin and C-peptide concentrations (means ؎ SE during morning hyperinsulinemic, euglycemic clamps (shaded area) and hyperinsulinemic stepped hypoglycemic clamps in nondiabetic FIG. 3. Sleep efficiency (percent of time asleep; means ؎ SE) at the 55- and hyperinsulinemic stepped hypoglycemic clamps and 45-mg/dl hypoglycemic steps on the sleep night in nondiabetic (8 ؍ subjects (A; n ؍ P .(4 ؍ and patients with type 1 diabetes (p; n (8 ؍ studied in the morning subjects (f; n ,(8 ؍ in patients with type 1 diabetes (B; n (0730–1230 h) while awake (E) and during the night (2100–0200 h) 0.0109 for diabetic vs. nondiabetic subjects in 45-mg/dl hypoglycemic while awake (F) and asleep (0000–0200 h; f). step.

DIABETES, VOL. 52, MAY 2003 1197 SLEEP AND HYPOGLYCEMIA

There was no significant effect of sleep on the cortisol response to hypoglycemia in the nondiabetic subjects (Fig. 8). In diabetic subjects, the plasma cortisol response to hypoglycemia at night was reduced (P ϭ 0.0069) during sleep compared with when the patients were awake (Fig. 8). Growth hormone. In the nondiabetic subjects, incre- ments in plasma growth hormone concentrations during hypoglycemia were similar in the morning and night awake states (Fig. 9). There was no significant effect of sleep on the plasma growth hormone response to hypo- glycemia (Fig. 9). In diabetic subjects, the increments in plasma growth hormone during hypoglycemia were greater (P ϭ 0.0197) in the night than in the morning awake state (Fig. 9). The growth hormone response to hypoglycemia was not reduced during sleep (Fig. 9). -FIG. 4. Plasma epinephrine concentrations (means ؎ SE) during Glucose infusion rate. The glucose infusion rates re morning hyperinsulinemic, euglycemic clamps (shaded area) and hy- perinsulinemic stepped hypoglycemic clamps in nondiabetic subjects quired to maintain the hypoglycemic steps were higher in studied in the the morning awake than in the night awake state in both (8 ؍ and patients with type 1 diabetes (B; n (8 ؍ A; n) morning (0730–1230) while awake (E) and during the night (2100– nondiabetic and diabetic subjects (P Ͻ 0.0001 and P ϭ 0200) while awake (F) and asleep (0000–0200; f). In diabetic sub- for night asleep vs. night awake state. 0.0156, respectively). In the nondiabetic subjects, the 0.0010 ؍ jects, P glucose infusion rate required to maintain the lower subjects, there were no glucagon responses to hypoglyce- glucose steps at night was higher (P ϭ 0.0022) when they mia under any of the study conditions (Fig. 6). were asleep than when they were awake (Fig. 10). In Pancreatic polypeptide. In the nondiabetic subjects, diabetic subjects, the required glucose infusion rates were increments in plasma pancreatic polypeptide concentra- similar when the subjects were awake and asleep (Fig. 10). tions during hypoglycemia were reduced (P ϭ 0.0003) Metabolic intermediates. In the nondiabetic subjects, during the night awake compared with the morning awake blood lactate (Table 5), serum nonesterified fatty acid state (Fig. 7). There was no effect of sleep on the pancre- (Table 6), blood ␤-hydroxybutyrate, and blood alanine atic polypeptide response (Fig. 7). In diabetic subjects, the (data not shown) concentrations were similar under all small pancreatic polypeptide responses to hypoglycemia hypoglycemic conditions. In the diabetic subjects, incre- were similar in the morning and night awake states (Fig. ments in blood lactate during hypoglycemia were reduced 7), but reduced (P ϭ 0.0034) in the night asleep state at night when the subjects were asleep (P ϭ 0.013) (Fig. 7). compared with in the morning when they were awake and Cortisol. Plasma cortisol levels at baseline were lower at tended to be reduced at night when the subjects were night than in the morning in both groups (Fig. 8). In the awake (P ϭ 0.0976) (Table 5). In addition, lactate levels nondiabetic subjects, increments in plasma cortisol con- were slightly higher during the asleep compared with the centrations during hypoglycemia, after being adjusted for awake study (P ϭ 0.0038) (Table 5). baseline differences, were greater (P ϭ 0.0005) in the night Heart rate and blood pressure. Heart rates were similar awake than in the morning awake state (Fig. 8). That was during hypoglycemia under all study conditions in both also the case (P ϭ 0.0384) in diabetic subjects (Fig. 8). groups (data not shown). Similarly, there were no differ- ences in systolic or diastolic blood pressures during hypoglycemia (data not shown).

DISCUSSION These data document markedly reduced awakening during hypoglycemia in patients with type 1 diabetes, a novel finding that could be attributed to diabetic patients’ re- duced sympathoadrenal responses to hypoglycemia during sleep, which were also documented by the present data. Sleep efficiency (the percent of time asleep) was threefold greater in the diabetic subjects than in the matched nondiabetic control subjects late in the 45-mg/dl hypogly- cemic steps; the diabetic subjects were asleep ϳ75% of the time, whereas the control subjects were awake ϳ75% of the time. Comparable data are not available from the study of Jones et al. (10), as they did not study nondiabetic FIG. 5. Plasma norepinephrine concentrations (means ؎ SE) during morning hyperinsulinemic, euglycemic clamps (shaded area) and hy- control subjects at night while awake. perinsulinemic stepped hypoglycemic clamps in nondiabetic subjects These data also document a reduced plasma epineph- -and hyperinsulinemic stepped hypoglycemic clamps in rine, and perhaps norepinephrine, response to hypoglyce (8 ؍ A; n) studied in the morning (8 ؍ patients with type 1 diabetes (B; n (0730–1230) while awake (E) and during the night (2100–0200) while mia during sleep in adults with uncomplicated type 1 awake (F) and asleep (0000–0200; f). diabetes, findings similar to those of Jones et al. (10) in

1198 DIABETES, VOL. 52, MAY 2003 S. BANARER AND P.E. CRYER

TABLE 3 Neurogenic (autonomic) symptom scores during hyperinsulinemic euglycemic and stepped hypoglycemic clamps in nondiabetic subjects and hyperinsulinemic stepped hypoglycemic clamps in patients with type 1 diabetes Nondiabetic Hypoglycemia Diabetic Hypoglycemia Night Night Night Night Time (min) Euglycemia Morning awake asleep Morning awake asleep –15 3 Ϯ 12Ϯ 22Ϯ 13Ϯ 11Ϯ 02Ϯ 14Ϯ 1 02Ϯ 11Ϯ 13Ϯ 13Ϯ 11Ϯ 02Ϯ 13Ϯ 1 30 2 Ϯ 11Ϯ 14Ϯ 23Ϯ 21Ϯ 02Ϯ 14Ϯ 1 60 3 Ϯ 12Ϯ 15Ϯ 23Ϯ 22Ϯ 13Ϯ 13Ϯ 1 90 3 Ϯ 13Ϯ 26Ϯ 33Ϯ 23Ϯ 13Ϯ 14Ϯ 1 120 3 Ϯ 13Ϯ 25Ϯ 23Ϯ 23Ϯ 13Ϯ 14Ϯ 1 150 4 Ϯ 13Ϯ 25Ϯ 35Ϯ 33Ϯ 13Ϯ 13Ϯ 1 180 4 Ϯ 14Ϯ 25Ϯ 3 — 3 Ϯ 15Ϯ 2 — 210 4 Ϯ 16Ϯ 28Ϯ 3 — 3 Ϯ 15Ϯ 1 — 240 5 Ϯ 27Ϯ 210Ϯ 4 — 4 Ϯ 15Ϯ 1 — 270 6 Ϯ 29Ϯ 214Ϯ 3 — 6 Ϯ 26Ϯ 2 — 300 4 Ϯ 112Ϯ 214Ϯ 2 — 8 Ϯ 25Ϯ 2 — Data are means Ϯ SE. Subjects were studied in the morning (0730–1230) and during the night (2100–0200) while awake and asleep (0000–0200). n ϭ 8 for both groups. adolescents with type 1 diabetes. A reduced plasma pan- polypeptide) responses to hypoglycemia during the sleep creatic polypeptide response to hypoglycemia in type 1 study in nondiabetic individuals. diabetic patients is also documented. Thus, taken to- Under all study conditions—morning or night, awake or gether, the data indicate that the autonomic nervous asleep—the diabetic subjects compared with matched system response to a given level of hypoglycemia is nondiabetic control subjects exhibited absent glucagon reduced during sleep in patients with type 1 diabetes. responses to hypoglycemia and reduced autonomic— Because arousal is a recognized adrenergic manifestation adrenomedullary (plasma epinephrine and norepineph- of the sympathoadrenal response to hypoglycemia (e.g., rine), sympathetic neural (neurogenic symptoms and the symptom nervous/anxious can be blocked by admin- plasma norepinephrine), and parasympathetic neural istration of catecholamine antagonists) (12), it is reason- (plasma pancreatic polypeptide)—responses to a given able to attribute the reduced awakening from sleep in the level of hypoglycemia, as expected (1). Although the loss patients to their reduced sympathoadrenal responses dur- of glucagon response in the diabetic subjects was abso- ing sleep. Our data and those of Jones et al. (10) also lute, autonomic responses could be elicited, but the gly- document a reduced adrenocortical cortisol response to cemic thresholds for those responses were shifted to hypoglycemia during sleep in the patients. However, in lower plasma glucose concentrations, again as expected contrast to the findings of Jones et al. (10), the present (1). The absent glucagon and reduced epinephrine re- data do not demonstrate significantly reduced plasma sponses were reflected biologically by the higher glucose epinephrine, norepinephrine, or cortisol (or pancreatic infusion rates required to maintain the lower hypoglyce-

TABLE 4 Neuroglycopenic symptom scores during hyperinsulinemic euglycemic and stepped hypoglycemic clamps in nondiabetic subjects and hyperinsulinemic stepped hypoglycemic clamps in patients with type 1 diabetes Nondiabetic Hypoglycemia Diabetic Hypoglycemia Night Night Night Night Time (min) Euglycemia Morning awake asleep Morning awake asleep –15 3 Ϯ 11Ϯ 01Ϯ 02Ϯ 12Ϯ 11Ϯ 13Ϯ 1 03Ϯ 01Ϯ 02Ϯ 12Ϯ 12Ϯ 11Ϯ 12Ϯ 0 30 3 Ϯ 11Ϯ 03Ϯ 12Ϯ 11Ϯ 11Ϯ 13Ϯ 0 60 3 Ϯ 11Ϯ 04Ϯ 23Ϯ 12Ϯ 12Ϯ 14Ϯ 1 90 3 Ϯ 12Ϯ 15Ϯ 24Ϯ 15Ϯ 12Ϯ 15Ϯ 1 120 3 Ϯ 12Ϯ 13Ϯ 13Ϯ 14Ϯ 12Ϯ 15Ϯ 1 150 4 Ϯ 12Ϯ 14Ϯ 25Ϯ 23Ϯ 14Ϯ 15Ϯ 1 180 3 Ϯ 12Ϯ 15Ϯ 2 — 3 Ϯ 14Ϯ 2 — 210 3 Ϯ 14Ϯ 28Ϯ 3 — 3 Ϯ 15Ϯ 2 — 240 4 Ϯ 16Ϯ 210Ϯ 3 — 4 Ϯ 17Ϯ 2 — 270 4 Ϯ 15Ϯ 214Ϯ 3 — 6 Ϯ 27Ϯ 2 — 300 3 Ϯ 111Ϯ 312Ϯ 2 — 7 Ϯ 27Ϯ 2 — Data are means Ϯ SE. Subjects were studied in the morning (0730–1230) and during the night (2100–0200) while awake and asleep (0000–0200). n ϭ 8 for both groups.

DIABETES, VOL. 52, MAY 2003 1199 SLEEP AND HYPOGLYCEMIA

FIG. 8. Plasma cortisol concentrations (means ؎ SE) during morning FIG. 6. Plasma glucagon concentrations (means ؎ SE) during morning hyperinsulinemic, euglycemic clamps (shaded area) and hyperinsuline- hyperinsulinemic, euglycemic clamps (shaded area) and hyperinsuline- (8 ؍ mic stepped hypoglycemic clamps in nondiabetic subjects (A; n (8 ؍ mic stepped hypoglycemic clamps in nondiabetic subjects (A; n and hyperinsulinemic stepped hypoglycemic clamps in patients with and hyperinsulinemic stepped hypoglycemic clamps in patients with studied in the morning (0730–1230) while (8 ؍ type 1 diabetes (B; n studied in the morning (0730–1230) while (8 ؍ type 1 diabetes (B; n awake (E) and during the night (2100–0200) while awake (F) and awake (E) and during the night (2100–0200) while awake (F) and for night 0.0005 ؍ asleep (0000–0200; f). In nondiabetic subjects, P asleep (0000–0200; f). for 0.0384 ؍ awake vs. morning awake state; in diabetic subjects, P .for night asleep vs 0.0069 ؍ night awake vs. morning awake state, P mic steps. The reduced sympathoadrenal (sympathetic night awake state. neural and adrenomedullary) responses were reflected by the reduced neurogenic (autonomic) symptom responses and norepinephrine responses, was similar in the morning to a given level of hypoglycemia in the diabetic compared and at night in the awake nondiabetic subjects; there were with the nondiabetic control subjects. no glucagon responses in the diabetic subjects, as ex- In general, subjects’ responses to hypoglycemia were pected. Baseline plasma cortisol concentrations were sub- remarkably similar in the morning and night awake states. stantially lower at night, also as expected. However, after Sympathoadrenal responses and the resultant neurogenic baseline adjustment, the cortisol responses to hypoglyce- symptom responses to hypoglycemia were similar in the mia were enhanced during the night in both the awake morning and night awake states in nondiabetic control control subjects and the awake diabetic subjects. Notably, subjects and, despite differences in the absolute values, in the glucose infusion rates required to maintain the hyper- the diabetic subjects. However, the parasympathetic neu- insulinemic glucose clamps were lower throughout (i.e., ral (plasma pancreatic polypeptide) response to hypogly- during euglycemia as well as hypoglycemia) at night cemia was reduced at night in the awake nondiabetic compared with in the morning in both groups. This finding subjects; the small pancreatic polypeptide response in the implies relative at night, a finding at awake diabetic subjects was not reduced significantly. The variance with an earlier report (24). Thus, aside from the glucagon response to hypoglycemia, like the epinephrine parasympathetic neural response, these data indicate that there is no diurnal variation per se in the physiological

(FIG. 7. Plasma pancreatic polypeptide concentrations (means ؎ SE during morning hyperinsulinemic, euglycemic clamps (shaded area) FIG. 9. Plasma growth hormone concentrations (means ؎ SE) during and hyperinsulinemic stepped hypoglycemic clamps in nondiabetic morning hyperinsulinemic, euglycemic clamps (shaded area) and hy- and hyperinsulinemic stepped hypoglycemic clamps perinsulinemic stepped hypoglycemic clamps in nondiabetic subjects (8 ؍ subjects (A; n and hyperinsulinemic stepped hypoglycemic clamps in (8 ؍ studied in the morning (A; n (8 ؍ in patients with type 1 diabetes (B; n studied in the morning (8 ؍ while awake (E) and during the night (2100–0200) while patients with type 1 diabetes (B; n (1230–0730) while awake (E) and during the night (2100–0200) while (1230–0730) ؍ awake (F) and asleep (0000–0200; f). In nondiabetic subjects, P 0.0197 ؍ for night awake vs. morning awake state; in diabetic subjects, awake (F) and asleep (0000–0200; f). In diabetic subjects, P 0.0003 .for night asleep vs. night awake states. for night awake vs. morning awake state 0.0034 ؍ P

1200 DIABETES, VOL. 52, MAY 2003 S. BANARER AND P.E. CRYER

and when they were ostensibly asleep. However, the epinephrine, norepinephrine, and pancreatic polypeptide responses appeared to be reduced at the 65- and 55-mg/dl hypoglycemic steps in the sleep compared with the awake nocturnal study. Notably, the significantly higher glucose infusion rates required to maintain the lower hypoglyce- mic steps during sleep suggest a biologically important difference. It is likely that the absence of an overall difference between the curves for the epinephrine, norepi- nephrine, and pancreatic polypeptide (and cortisol) re- sponses was the result of the lower sleep efficiency in the nondiabetic control subjects. They were asleep only ϳ25% of the time late in the lowest hypoglycemic step. Thus, the apparent lack of an impact of sleep in the nondiabetic control subjects may well have been the result of the fact .FIG. 10. Glucose infusion rates (means ؎ SE) during morning hyper- that, for the most part, they were not asleep insulinemic, euglycemic clamps (shaded area) and hyperinsulinemic The significantly higher glucose infusion rates required and to maintain the lower hypoglycemic steps during sleep in (8 ؍ stepped hypoglycemic clamps in nondiabetic subjects (A; n hyperinsulinemic stepped hypoglycemic clamps in patients with type 1 studied in the morning (0730–1230) while awake the nondiabetic subjects suggest a biological impact of our (8 ؍ diabetes (B; n (E) and during the night (2100–0200) while awake (F) and asleep interpretation that their epinephrine responses were re- (0000–0200; f). In nondiabetic subjects, P < 0.0001 for morning duced earlier during the hypoglycemic clamps under that for night asleep vs. night 0.0022 ؍ awake vs. night awake state, P for diabetic subjects for condition. The required glucose infusion rates were higher 0.0156 ؍ awake state; in diabetic subjects, P morning awake vs. night awake state. during hypoglycemia in the diabetic subjects, who had absent glucagon and reduced epinephrine responses to responses to hypoglycemia. Because the parasympathetic hypoglycemia, than in the nondiabetic control subjects, as response has no known direct role in defense against expected. However, there was no difference in the glucose developing hypoglycemia (1), it follows that there is no infusion rates required to maintain the lower hypoglyce- diurnal variation, independent of sleep, in the physiology mic steps in the diabetic subjects during sleep, despite the of glucose counterregulation. subjects’ further reduced epinephrine responses. Thus it Taken at face value, the impact of sleep appeared to appears that this measure was not sufficiently sensitive to differ in the nondiabetic control and diabetic subjects. As document the anticipated effect in the diabetic subjects, noted earlier, the adrenomedullary (plasma epinephrine) who had substantially impaired glucose counterregulation and parasympathetic neural (plasma pancreatic polypep- under all three study conditions. tide) responses to hypoglycemia of the diabetic subjects The reduced cortisol response to hypoglycemia during were reduced significantly, and the plasma norepinephrine sleep in diabetic subjects is relevant to the suggestion that response to hypoglycemia appeared to be reduced, during it is the cortisol response to prior hypoglycemia that the night asleep compared with the night awake state. In mediates reduced autonomic (including epinephrine) and contrast, during the nocturnal studies in the nondiabetic symptomatic responses to subsequent hypoglycemia (25– control subjects, the plasma epinephrine, norepinephrine, 26), that is, that cortisol is the mediator of hypoglycemia- and pancreatic polypeptide responses to hypoglycemia associated autonomic failure in diabetes (1,6). Nocturnal were statistically similar when the subjects were awake hypoglycemia reduces the autonomic (including epineph-

TABLE 5 Blood lactate concentrations (␮mol/l) during hyperinsulinemic euglycemic and stepped hypoglycemic clamps in nondiabetic subjects and hyperinsulinemic stepped hypoglycemic clamps in patients with type 1 diabetes Nondiabetic Hypoglycemia Diabetic Hypoglycemia Time (min) Euglycemia Morning Night awake Night asleep Morning Night awake* Night asleep –15 715 Ϯ 253 854 Ϯ 82 686 Ϯ 52 597 Ϯ 66 630 Ϯ 127 723 Ϯ 132* 624 Ϯ 33 0 391 Ϯ 138 717 Ϯ 66 632 Ϯ 23 528 Ϯ 75 597 Ϯ 95 678 Ϯ 116* 571 Ϯ 33 30 312 Ϯ 110 905 Ϯ 58 818 Ϯ 15 814 Ϯ 80 774 Ϯ 87 711 Ϯ 83* 775 Ϯ 39† 60 220 Ϯ 78 1,354 Ϯ 110 1,120 Ϯ 110 1,032 Ϯ 120 984 Ϯ 98 772 Ϯ 97* 911 Ϯ 81 90 249 Ϯ 88 1,164 Ϯ 58 1,007 Ϯ 87 1,008 Ϯ 93 956 Ϯ 126 731 Ϯ 73* 895 Ϯ 65 120 511 Ϯ 181 1,099 Ϯ 73 832 Ϯ 67 968 Ϯ 92 1,015 Ϯ 93 709 Ϯ 74* 902 Ϯ 87 150 471 Ϯ 166 1,108 Ϯ 132 771 Ϯ 62 896 Ϯ 83 975 Ϯ 115 671 Ϯ 66* 805 Ϯ 87 180 353 Ϯ 125 1,114 Ϯ 107 801 Ϯ 110 814 Ϯ 85 965 Ϯ 116 619 Ϯ 76* 773 Ϯ 82 210 249 Ϯ 88 1,184 Ϯ 125 1,007 Ϯ 121 896 Ϯ 112 1,007 Ϯ 110 618 Ϯ 71 736 Ϯ 65 240 298 Ϯ 105 1,207 Ϯ 159 1,193 Ϯ 134 948 Ϯ 121 921 Ϯ 118 667 Ϯ 73 666 Ϯ 59 270 288 Ϯ 109 1,427 Ϯ 172 1,491 Ϯ 185 1,192 Ϯ 190 930 Ϯ 77 770 Ϯ 69* 622 Ϯ 61 300 373 Ϯ 131 1,632 Ϯ 223 1,810 Ϯ 263 1,564 Ϯ 283 1,037 Ϯ 102 665 Ϯ 88 668 Ϯ 67 Data are means Ϯ SE. Subjects were studied in the morning (0730–1230) and during the night (2100–0200) while awake and asleep (0000–0020). n ϭ 8 for both groups. *P ϭ 0.0976 vs. morning; †P ϭ 0.0130 vs. morning, P ϭ 0.0038 vs. night awake.

DIABETES, VOL. 52, MAY 2003 1201 SLEEP AND HYPOGLYCEMIA

TABLE 6 Serum nonesterified fatty acid concentrations (␮mol/l) during hyperinsulinemic euglycemic and stepped hypoglycemic clamps in nondiabetic subjects and hyperinsulinemic stepped hypoglycemic clamps in patients with type 1 diabetes Nondiabetic Hypoglycemia Diabetic Hypoglycemia Time (min) Euglycemia Morning Night awake Night asleep Morning Night awake Night asleep –15 470 Ϯ 52 539 Ϯ 50 709 Ϯ 90 655 Ϯ 61 290 Ϯ 58 377 Ϯ 102 385 Ϯ 68 0 458 Ϯ 47 506 Ϯ 33 718 Ϯ 73 649 Ϯ 53 302 Ϯ 44 397 Ϯ 106 371 Ϯ 69 30 211 Ϯ 59 255 Ϯ 65 361 Ϯ 80 306 Ϯ 66 113 Ϯ 14 156 Ϯ 47 133 Ϯ 22 60 97 Ϯ 24 86 Ϯ 21 177 Ϯ 48 165 Ϯ 66 63 Ϯ 11 89 Ϯ 20 96 Ϯ 13 90 85 Ϯ 20 83 Ϯ 21 198 Ϯ 54 103 Ϯ 21 60 Ϯ 10 72 Ϯ 21 77 Ϯ 8 120 83 Ϯ 18 75 Ϯ 20 127 Ϯ 25 85 Ϯ 21 65 Ϯ 10 62 Ϯ 19 66 Ϯ 11 150 82 Ϯ 21 62 Ϯ 14 103 Ϯ 22 90 Ϯ 23 54 Ϯ 857Ϯ 18 84 Ϯ 11 180 66 Ϯ 16 58 Ϯ 16 99 Ϯ 18 87 Ϯ 23 56 Ϯ 958Ϯ 19 73 Ϯ 10 210 62 Ϯ 13 68 Ϯ 17 96 Ϯ 18 82 Ϯ 20 54 Ϯ 845Ϯ 10 47 Ϯ 7 240 62 Ϯ 12 68 Ϯ 22 100 Ϯ 22 67 Ϯ 19 54 Ϯ 14 55 Ϯ 14 43 Ϯ 11 270 53 Ϯ 12 86 Ϯ 25 132 Ϯ 27 104 Ϯ 25 61 Ϯ 14 54 Ϯ 15 54 Ϯ 5 300 53 Ϯ 12 103 Ϯ 29 154 Ϯ 29 133 Ϯ 32 74 Ϯ 23 81 Ϯ 23 52 Ϯ 12 Data are means Ϯ SE. Subjects were studied in the morning (0730–1230) and during the night (2100–0200) while awake and asleep (0000–0200). n ϭ 8 for both groups. rine) and symptomatic responses to hypoglycemia the REFERENCES next morning in patients with type 1 diabetes (27). To the 1. Cryer PE: Hypoglycaemia: The limiting factor in the glycaemic manage- extent that those patients slept during hypoglycemia, the ment of type I and type II diabetes. Diabetologia 45:937–948, 2002 present data suggest that the cortisol levels were not 2. The Diabetes Control and Complications Trial Research Group: The effect elevated during the nocturnal hypoglycemia. If so, the of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl reduced responses to hypoglycemia the next morning J Med 329:977–986, 1993 cannot be attributed to cortisol elevations during the 3. The United Kingdom Prospective Diabetes Study Group: Intensive blood- earlier nocturnal hypoglycemia. glucose control with sulphonylureas or insulin compared with conven- The finding of a reduced cortisol response to hypogly- tional treatment and risk of complications in patients with . cemia during sleep in diabetic subjects may also be Lancet 352:837–853, 1998 relevant to the fact that Bendtson et al. (9) did not find 4. Dagogo-Jack SE, Craft S, Cryer PE: Hypoglycemia-associated autonomic failure in insulin-dependent diabetes mellitus: recent antecedent hypogly- reduced sympathoadrenal responses to hypoglycemia in cemia reduces autonomic responses to, symptoms of, and defense against their nocturnal study. Given their brisk cortisol responses, subsequent hypoglycemia. J Clin Invest 91:819–828, 1993 it is likely that the patients studied were to a large extent 5. Segel SA, Paramore DS, Cryer PE: Hypoglycemia-associated autonomic awake rather than asleep. failure in advanced type 2 diabetes. Diabetes 51:724–733, 2002 We conclude that autonomic responses to hypoglycemia 6. Cryer PE: Hypoglycemia-associated autonomic failure in diabetes. Am J are further reduced during sleep in patients with type 1 Physiol 281:E1115ϪE121, 2001 7. The Diabetes Control and Complications Control Research Group: Epide- diabetes. Probably because of their reduced sympathoad- miology of severe hypoglycemia in the diabetes control and complications renal responses, these patients are substantially less likely trial. Am J Med 90:450–459, 1991 to be awakened by hypoglycemia. Thus sleep should be 8. The Diabetes Control and Complications Trial Research Group: Hypogly- added to recent antecedent hypoglycemia as a cause of cemia in the Diabetes Control and Complications Trial. Diabetes 46:271– hypoglycemia-associated autonomic failure in diabetes 286, 1997 (1,6). This sleep-related hypoglycemia-associated auto- 9. Bendtson I, Rosenfalck AM, Binder C: Nocturnal versus diurnal hormonal nomic failure, which impairs both physiological and ulti- counterregulation to hypoglycemia in type 1 (insulin-dependent) diabetes patients. Acta Endocrinol 128:109–115, 1993 mately (because of the lack of awakening) behavioral 10. Jones TW, Porter P, Sherwin RS, Davis EA, O’Leary P, Frazer F, Byrne G, defenses against developing hypoglycemia, in the context Stick S, Tamborlane WV: Decreased epinephrine responses to hypoglyce- of imperfect insulin replacement, likely explains the high mia during sleep. N Engl J Med 338:1657–1662, 1998 frequency of nocturnal hypoglycemia in type 1 diabetes. 11. Schwartz NS, Clutter WE, Shah SD, Cryer PE: The glycemic thresholds for activation of glucose counterregulatory systems are higher than the threshold for symptoms. J Clin Invest 79:777–781, 1987 ACKNOWLEDGMENTS 12. Towler DA, Havlin CE, Craft S, Cryer P: Mechanisms of awareness of This work was supported, in part, by a Clinical Research hypoglycemia: perception of neurogenic (predominantly cholinergic) rather than neuroglycopenic symptoms. Diabetes 42:1791–1798, 1993 Grant from the American Diabetes Association and by U.S. 13. Rechtschaffen A, Kales A (Eds.): A Manual of Standardized Terminology, Public Health Service Grants R37-DK-27085, M01-RR- Techniques and Scoring System for Sleep Stages in Human Subjects. Los 00036, and P60-DK-20579. Angeles,: UCLA Brain Information Service/Brain Research Institute, 1968 We acknowledge the technical assistance of Krishan 14. Kuzuya H, Blix PM, Horwitz DL, Steiner DF, Rubenstein AH: Determination Jethi, Cornell Blake, Joy Brothers, Michael Morris, and of free and total insulin and C-peptide in insulin-treated diabetics. Diabetes Sharon O’Neill; the assistance of the nursing staff at 26:22–29, 1977 15. Ensinck J: Immunoassays for glucagon. In Handbook of Experimental Washington University General Clinical Research Center; Pharmacology. Vol 66. Lefebrve P, Ed. New York, Springer Verlag, 1983, p. and the assistance of Karen Muehlhauser in the prepara- 203–221 tion of this manuscript. 16. Gingerich RL, Lacy PE, Chance RE, Johnson MG: Regional pancreatic

1202 DIABETES, VOL. 52, MAY 2003 S. BANARER AND P.E. CRYER

concentration and in-vitro secretion of canine pancreatic polypeptide, 23. Cahill GF Jr, Herrera MG, Morgan AP, Soeldner JS, Steinke J, Levy PL, insulin, and glucagon. Diabetes 27:96–101, 1978 Reichard GA Jr, Kipnis DM: Hormone-fuel interrelationships during fast- 17. Schalch D, Parker M: A sensitive double antibody radioimmunoassay for ing. J Clin Invest 45:1751–1769, 1966 growth hormone in plasma. Nature 703:1141–1142, 1964 24. Perriello G, DeFeo P, Torlone E, Fanelli C, Santuesanio F, Brunetti P, Bolli 18. Farmer RW, Pierce CE: Plasma cortisol determination: radioimmunoassay GB: The dawn phenomenon in type 1 (insulin dependent) diabetes and competitive protein binding compared. Clin Chem 20:411–414, 1974 mellitus: magnitude, frequency, variability and dependence on glucose 19. Shah SD, Clutter WE, Cryer PE: External and internal standards in the counterregulation and insulin sensitivity. Diabetes 34:21–28, 1991 single-isotope derivative (radioenzymatic) measurement of plasma norepi- 25. Davis SN, Shavers C, Costa F, Mosqueda-Garcia R: Role of cortisol in the nephrine and epinephrine. J Lab Clin Med 106:624–629, 1985 pathogenesis of deficient counterregulation after antecedent hypoglycemia 20. Hosaka K, Kijuchi T, Mitsuhida N, Kawaguchi A: A new colorimetric in normal humans. J Clin Invest 98:680–691, 1996 method for the determination of free fatty acids with acyl-CoA synthetase 26. Davis SN, Shavers C, Davis B, Costa F: Prevention of an increase in plasma and acyl-CoA oxidase. J Biochem (Tokyo) 89:1799–1803, 1981 21. Pinter J, Hayaski J, Watson J: Enzymatic assay of glycerol, dihydroxyac- cortisol during hypoglycemia preserves subsequent counterregulatory etone and glyceraldehyde. Arch Biochem Biophys 121:404–414, 1967 responses. J Clin Invest 100:429–438, 1997 22. Lowry O, Passoneau J, Hasselberger F, Schultz D: Effect of ischemia on 27. Fanelli CG, Paramore DS, Hershey T, Terkamp C, Ovalle F, Craft S, Cryer known substrates and co-factors of the glycolytic pathway of the brain. PE: Impact of nocturnal hypoglycemia on hypoglycemic cognitive dysfunc- J Biol Chem 239:18–30, 1964 tion in type 1 diabetes mellitus. Diabetes 47:1920–1927, 1998

DIABETES, VOL. 52, MAY 2003 1203