Published ahead of print on March 24, 2004, doi:10.1164/rccm.200308-1200OC
© 2004 American Thoracic Society
Hypoxia Causes Glucose Intolerance in HumansDepartments of Internal Medicine I, Anaesthesia, Psychiatry and Psychotherapy, and Neuroendocrinology, University of Luebeck, Luebeck, Germany Correspondence and requests for reprints should be addressed to Kerstin M. Oltmanns, M.D., Medical Clinic I, 23a, University of Luebeck, Ratzeburger Allee 160, 23538 Luebeck, Germany. E-mail: Oltmanns{at}medinf.mu-luebeck.de
Hypoxic respiratory diseases are frequently accompanied by glucose intolerance. We examined whether hypoxia is a cause of glucose intolerance in healthy subjects. In a double-blind within-subject crossover design, hypoxic versus normoxic conditions were induced in 14 healthy men for 30 minutes by decreasing oxygen saturation to 75% (versus 96% in control subjects) under the conditions of a euglycemic clamp. The rate of dextrose infusion needed to maintain stable blood glucose levels was monitored. Neurohormonal stress response was evaluated by measuring catecholamine and cortisol concentrations as well as cardiovascular parameters, and symptoms of anxiety. To differentiate between the effects of stress hormonal response, and hypoxia itself, on glucose intolerance, we performed hypoglycemic clamps as a nonspecific control. We found a significant decrease in dextrose infusion rate over a period of 150 minutes after the start of hypoxia (p < 0.01). Hypoxia also increased plasma epinephrine concentration (p < 0.01), heart rate (p < 0.01), and symptoms of anxiety (p < 0.05), whereas the other parameters remained unaffected. Glucose intolerance was closely comparable between hypoxic and hypoglycemic conditions (p < 0.9) despite clear differences in stress hormonal responses. Hypoxia acutely causes glucose intolerance. One of the factors mediating this effect could be an elevated release of epinephrine.
Key Words: catecholamines • chronic hypoxic diseases • clamp • heart rate • human Glucose intolerance is a basic clinical feature of the metabolic syndrome whose growing prevalence has led to an increase in cardiovascular morbidity and lethality in industrial countries. In respiratory diseases such as chronic pulmonary disease or sleep disorders with respiratory manifestations such as obstructive sleep apnea (OSA), glucose intolerance is frequently observed. Although the mechanism is unknown, glucose intolerance in respiratory diseases is commonly suspected to result from frequent and persistent hypoxic conditions (1–4). In OSA, glucose intolerance correlates positively with severity of the disease (5–7) and therapy with continuous positive airway pressure (CPAP) improves glucose tolerance significantly (1, 8). Obesity is the main risk factor for glucose intolerance and diabetes, but OSA is associated with glucose intolerance independent of body mass index (7, 9–13). Moreover, studies have demonstrated that severe OSA substantially adds to this risk (14). However, OSA is also characterized by various comorbidities and substantial sleep debt, in addition to chronically occurring hypoxia, which renders the view of hypoxia as the primary and sole cause of glucose intolerance questionable. Spiegel and coworkers demonstrated that isolated sleep debt without hypoxia induces significant glucose intolerance in healthy young men (15). Similar to OSA, glucose intolerance can be observed in hypoxemic patients with chronic pulmonary disease (4) or cystic fibrosis (16, 17). Although all these observations indicate a close association between hypoxia and the emergence of glucose intolerance, experimental evidence of a causative role for hypoxia in this metabolic dysfunction is surprisingly lacking. We addressed this issue by testing the effects of hypoxia versus normoxia, using the "gold standard" method to measure glucose tolerance: the euglycemic insulin clamp technique (18). By infusion of exogenous glucose, endogenous production of glucose is completely suppressed. Therefore, it is possible to conclude from a decreased glucose infusion rate that glucose tolerance is decreased and vice versa. In parallel, we measured concentrations of circulating catecholamines and cortisol as well as heart rate, blood pressure, and self-rated symptoms of anxiety. We found a significant decrease in glucose infusion rate during hypoxia in conjunction with increased epinephrine concentration and symptoms of anxiety. These results leave unresolved whether glucose intolerance was secondary to hypoxia or, rather, anxiety and increased epinephrine. On this background as a nonspecific control intervention, we performed hypoglycemic clamp experiments. Hypoglycemia is well known to induce glucose intolerance as well as pronounced increases in catecholamine and cortisol concentrations (19–21). Our rationale: if glucose intolerance occurred solely due to the rise in stress hormones on hypoxia, then we would expect much stronger effects on glucose metabolism after hypoglycemia than after hypoxia. In fact, despite clear differences in stress hormonal responses, glucose intolerance was closely comparable between hypoxic and hypoglycemic conditions.
Subjects Fourteen healthy white men (age [mean ± SD], 24.3 ± 3.5 years) with a body mass index less than 25 kg/m2 (mean ± SD, 22.1 ± 1.4 kg/m2) participated in the experiments. Exclusion criteria included respiratory diseases, chronic or acute illness, anxiety disorders, alcohol or drug abuse, smoking, competitive sports, exceptional physical or psychologic stress, and current medication of any kind. Each volunteer gave written informed consent, and the study was approved by the local ethics committee.
Experimental Design and Procedure After 3 hours, hypoxia was induced for 30 minutes by decreasing oxygen saturation to 75% (versus 96% in the control condition). Heart rate was determined every 5 minutes and blood pressure was determined every 15 minutes. Blood samples for determination of stress hormone response were collected every 7.5 minutes during the intervention, starting with induction of hypoxia (versus normoxia). After 30 minutes, oxygen saturation was quickly normalized and the Acute Panic Inventory was repeated, with reference to the 30-minute intervention period. During the remaining 2 hours, blood samples were collected every 30 minutes. Heart rate and blood pressure were monitored hourly. A detailed description of the procedure and assays can be found in the online supplement.
Hypoglycemic Clamp Procedure
Statistical Analysis
Respiratory Parameters, Serum Insulin, and Plasma Glucose Oxygen saturation decreased in the hypoxic session to a plateau averaging 74.3 ± 2.4%. In the normoxic control session, oxygen saturation averaged 98.3 ± 1.7%, and in the hypoglycemic session it was 97.9 ± 1.2% (Figure 1A) . Tidal volume, respiratory rate, and end-tidal volume were comparable in hypoxia and normoxia whereas arterial PCO2 was lower during hypoxia (p = 0.045; Table 1) . Serum insulin did not differ between the three conditions (Figure 1B). Plasma glucose was equal in the hypoxic and the normoxic control sessions whereas values decreased to a mean of 2.65 mmol/L under hypoglycemic conditions (Figure 1B).
Dextrose Infusion Rates Dextrose infusion rates, as expected, distinctly rose during the first 3 hours of euglycemic clamping before the hypoxemic intervention, with this rise being closely comparable for both conditions (Figure 2A) . The average infusion rate before hypoxic intervention was 269.3 ± 11.4 and 266.4 ± 15.8 ml/hour for the corresponding interval of normoxic control condition (p = 0.81). Induction of hypoxia, however, induced a distinct decrease in the dextrose infusion rate, indicating reduced glucose tolerance as compared with the normoxic control condition (p = 0.04 for the session effect in an analysis of variance of the average infusion rate during a 50-minute interval after onset of hypoxia; p = 0.01 for the session-by-time interaction; Figure 2A). The effect developed gradually after the start of hypoxia, reached its peak 20 minutes after the hypoxic period had ended (i.e., 50 minutes after onset of hypoxia), and did not decrease before 180 minutes after hypoxia induction (p = 0.01 for the session effect for average infusion rates 50–180 minutes after hypoxia induction; p = 0.02 for the session-by-time interaction; Figure 2). Thereafter, differences in dextrose infusion rate between conditions declined.
In the hypoglycemic condition, dextrose infusion rates after hypoglycemia were decreased compared with the euglycemic normoxic control condition (p = 0.04 for the session effect for average infusion rates 60–180 minutes after the beginning of the hypoglycemic plateau; Figure 2B). A comparison between hypoxic and hypoglycemic conditions during the same time period revealed no differences (p = 0.19 for the session effect and p = 0.97 for the session-by-time interaction; Figure 2B).
Epinephrine, Norepinephrine, and Cortisol
Heart Rate, Blood Pressure, and Anxiety Symptoms Hypoxia induced an acute elevation of heart rate during the 30-minute intervention period, averaging 19 beats/minute (p = 0.001; Table 3) . In contrast, systolic and diastolic blood pressure were not affected by hypoxia (p = 0.19 and p = 0.86, respectively; Table 3). Comparison of hypoglycemia and the euglycemic normoxic control condition revealed no differences regarding heart rate and systolic blood pressure (p = 0.998 and p = 0.074, respectively; Table 3). Diastolic blood pressure was lower in the hypoglycemic condition than in the control condition (p = 0.004; Table 3). Acute Panic Inventory scores were increased after the period of intervention as compared with scores before intervention (p < 0.001, for the overall effect of time). This increase was stronger after hypoxia (from 4 ± 2 to 11 ± 7) than after normoxia (from 3 ± 2 to 6 ± 3; p = 0.03 for session-by-time interaction). In addition, the increase after hypoglycemia as compared with the control condition was even stronger (p = 0.02 for session-by-time interaction).
The present results demonstrate that hypoxia acutely induces glucose intolerance. Although it is frequently observed that diseases characterized by chronic hypoxic episodes are associated with glucose intolerance (1–14, 16, 17), the available literature is not conclusive on this association. So far it has been unclear whether glucose intolerance is a result of hypoxia itself or of other factors influencing glucose metabolism such as sleep debt or the various comorbidities frequently associated with these diseases. Although providing evidence of a causative role for hypoxia in acutely elevating glucose intolerance, it remains to be carefully considered to what extent our findings after acute experimental short-term hypoxia can be generalized to the typical pathologic conditions in which frequent hypoxic episodes are encountered over long periods. High-altitude and in vitro studies have been used as alternative models to test metabolic consequences of hypoxia. Several studies investigated acclimatization to high altitude with lowered oxygen content. Congruent with the present findings, a short, 2-day stay at 4,550 m induced marked glucose intolerance in healthy men (25) as well as women (26). With more prolonged stays, this effect seems to disappear (27). However, whether extended exposure to high altitude more closely reflects the conditions present during the respiratory diseases of interest here must be questioned. For example, in OSA, frequent and discrete episodes of hypoxia occur selectively during sleep, whereas the effects of high altitude are usually investigated for the total 24-hour cycle, which also includes periods of physical activity. During exercise, glucose is transported into the skeletal muscle to a substantial extent independent of insulin, which confounds any estimate of glucose tolerance under these conditions. It is similarly difficult to control for confounding exercise in an animal model. Polotsky and coworkers found in lean and leptin-deficient mice an improvement in glucose tolerance after exposure to intermittent hypoxia for 5 days as compared with normoxic control (28). A potential decrease in glucose tolerance during hypoxia may be mimicked by increased glucose uptake based on stress-induced agitation. After intermittent exposure to hypoxia for 12 weeks in the same study, glucose intolerance increased (28), probably indicating an adaptation to hypoxic stress followed by less agitation. In an in vitro approach, Azevedo and coworkers (29) reported signs of a hypoxia-induced increase in glucose transport in isolated strips of human skeletal muscle. However, in vitro investigations of glucose metabolism under hypoxia are subject to even stronger limitations due to their disconnection from a number of hormonal (e.g., catecholamines, cortisol) and neuronal inputs that in vivo essentially contribute to glucose regulation. We are aware that a single 30-minute period of continuous hypoxia in an awake subject cannot imitate the hypoxic conditions present in OSA because the disease is classically characterized by hypoxia/reoxygenation occurring during sleep, with each discrete episode of severe desaturation lasting for only seconds to minutes, and not 20–30 minutes. Nevertheless, this is the first study investigating effects of a decrease in oxygen saturation to 75% within 8 minutes in resting healthy humans, largely excluding confounding factors such as exercise. In response to a stressor, glucose metabolism becomes subject to control by the two major neurohormonal stress systems, that is, the sympathoadrenal system and the hypothalamus–pituitary–adrenal axis. Both catecholamines and cortisol, via increasing hepatic glucose output and decreasing muscular glucose uptake (30, 31), enhance blood glucose concentration, with the actions of catecholamines, mainly of epinephrine, being faster than those of cortisol (32, 33). The present data provide hints at a differential role played by these stress hormones in the mediation of hypoxia-induced glucose intolerance, because experimental hypoxia induced an immediate increase selectively in plasma levels of epinephrine. Levels of norepinephrine and cortisol did not substantially change on hypoxia within the 3-hour monitoring interval. The pattern of changes in catecholamine levels with only epinephrine responding to hypoxia, fits well with observations in studies of high altitude. In these studies, epinephrine concentrations rose sharply at short-term exposure to high altitude whereas concentrations of norepinephrine rose with a substantial delay, with significant elevations occurring not until after 2 to 3 days of altitude exposure (26, 34–36). The idea of an immediate effect of hypoxia on epinephrine and a delayed effect of persisting hypoxia on plasma levels of norepinephrine is likewise supported by the observation in patients with OSA of elevated levels of epinephrine during nocturnal sleep as measured in urinary excretion (37). Studies with measurements limited to daytime, however, did not reveal increased levels of epinephrine (38–41). Rather, these studies report an increase in plasma and urinary concentrations of norepinephrine in patients with OSA, as compared with healthy control subjects, which decreased substantially (by 23–50%) after extensive CPAP therapy (40, 41). Thus, the short-lasting increase in epinephrine found in the current study, exceeding the hypoxic interval by no more than 20 minutes, fits well with past failures to find altered levels of epinephrine in patients with OSA during daytime measurements. In addition, the lack of a discrete cortisol response to hypoxia agrees with previous data, although the hypothalamus–pituitary–adrenal system has not been thoroughly investigated in this context. Several studies examining the effects of acute hypoxia at rest and during exercise failed to reveal any changes in cortisol levels at rest during hypoxia or shortly afterward (42–44). Data on cortisol are likewise rare in studies of hypoxic diseases. However, there has been no evidence in patients with chronic hypoxic pulmonary disease (4, 45), or cystic fibrosis or OSA (46–50), of altered cortisol concentration in comparison with healthy control subjects, with the exception of a single study (51) reporting elevated cortisol concentrations in patients with OSA. Unfortunately, in this study, patients and control subjects were not matched for body mass index, which is known to correlate with cortisol concentration (52, 53). Overall, the data lend themselves to conclude that the pituitary–adrenal stress system is not essentially involved in mediating glucose intolerance after acute or prolonged hypoxia. We measured heart rate, blood pressure, and subjective anxiety as indicators of autonomic nervous system arousal and of a psychological arousal response to the hypoxic stress. A hypoxically induced increase in heart rate is well documented in previous studies and was confirmed here (34, 54, 55). Regarding blood pressure, which in the current study remained unaffected by hypoxia, previous results have been inconsistent. Depending on the experimental design and time of measurement, some studies report acute increases primarily in diastolic pressure after hypoxia, whereas others do not. In one study of healthy men, in which acute hypoxic conditions at altitudes between 4,000 and 6,000 m were simulated, blood pressure did not change (54), whereas others found increased systolic and mean arterial blood pressure during prolonged periods of hypoxia (34, 55, 56). In patients with OSA, there is increasing evidence of an association between the disease and daytime hypertension (57, 58) that can be reduced by CPAP therapy, particularly in patients with nocturnal desaturation (58). One possibility is that the blood pressure response to hypoxia, like the norepinephrine response, develops with some delay or requires extended periods of hypoxia. This would be consistent with our failure to observe a systematic alteration in blood pressure after a single acute period of hypoxia, as well as with previous observations of increased daytime blood pressure after nocturnal hypoxia (59). We measured the state of anxiety to see whether oxygen desaturation to 75% is intense enough to cause psychologic arousal. Anxiety levels increased during hypoxia, which agrees with a previous study (60), but remained considerably lower than what is typically seen during panic attacks (22). Increased anxiety at the subjective level corresponds well with the concurrent increases in heart rate and epinephrine release forming parts of the immediate psychophysiologic stress response to hypoxia. Considering the changes in the various stress-related parameters, activation of the sympathoadrenal release of epinephrine appears to be the only stress-related mechanism contributing to hypoxia-induced glucose intolerance, probably via increasing hepatic glucose output and decreasing muscular glucose uptake (31). This view is supported by the observation that patients with OSA show enhanced sympathetic activity and decreased glucose tolerance. These alterations are both regressive after CPAP therapy (1, 8, 61, 62). However, although an increase in epinephrine seems to contribute to hypoxia-induced glucose intolerance, it cannot be the sole factor. To investigate whether glucose intolerance is induced by hypoxia itself or rather by anxiety and increased epinephrine, we performed additional hypoglycemic experiments, interventions that are well known to cause glucose intolerance (19–21). For best comparison, we conducted the experiments with the subjects from the hypoxic/control condition. If alone the rise in epinephrine was causative for glucose intolerance during hypoxia, we would have expected much stronger effects on glucose metabolism after the overwhelming increase in stress hormones during hypoglycemia. In fact, hypoglycemia did induce strong increases in catecholamine and cortisol concentrations. But there was no difference in glucose intolerance between hypoxic and hypoglycemic conditions. This lets us conclude that hypoxia itself probably has an additive effect on glucose intolerance that cannot be explained by the small increase in epinephrine. We can only speculate about the mechanism of hypoxia-induced glucose intolerance. Because this effect does not seem to be caused solely by an increase in circulating catecholamines, hypoxia-induced glucose intolerance may possibly be mediated by neural–adrenergic pathways. However, hypoxia in addition has direct peripheral effects that could participate in glucose regulation as well, and were not measured here. Thus, ATP-sensitive K+ channels are expressed in skeletal muscle tissue, which senses glucose levels (63, 64). A hypoxia-induced decrease in ATP levels and adrenergic input could open these K+ channels, thereby decreasing muscle glucose uptake. Although these and further mechanisms of hypoxia-induced glucose intolerance need further clarification, our results underscore the importance to control for early signs of disordered glucose metabolism in hypoxic disease. In contrast to our subjects, the majority of patients with OSA are overweight or obese. Considering that CPAP therapy (i.e., better oxygenation) improves glucose metabolism independent of weight loss (8), we assume that our findings may also be assigned to overweight subjects. However, our results suggest that the presence of hypoxic diseases should be evaluated as a possible pathogenic factor in patients suffering from metabolic disorders such as Type 2 diabetes and metabolic syndrome.
The authors thank Christiane Zinke for expert and invaluable laboratory assistance, Anja Otterbein for organizational work, and Dr. Mike Harnish as well as Dr. Lisa Marshall for language advice.
This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Conflict of Interest Statement: K.M.O. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; H.G. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; S.R. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; B.S does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; S.R. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; U.S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; J.B. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; H.L.F. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; A.P. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form August 29, 2003; accepted in final form March 22, 2004
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
