in this issue of the Journal of Applied Physiology, Tamisier and colleagues (20) describe a novel method for studying the effects of chronic intermittent hypoxia (CIH) in healthy human subjects. To investigate the effects of CIH on sleep, ventilatory control, and blood pressure, the authors make clever use of commercially available altitude tents to mimic the cyclical arterial oxygen desaturations-resaturations of sleep apnea. Young healthy men and women slept in the tents with a fraction of inspired oxygen (FiO2) of 13% for either 14 (Grenoble, France) or 28 (Boston, MA) consecutive nights. At 2-min intervals throughout the night, supplemental oxygen was administered for 15 s via a nasal cannula. The oxygen flow rate was adjusted for each subject so that the difference in nadir and peak arterial oxygen saturation (SaO2) was 10% (mean values: 85–95%). These procedures resulted in central hypopneas (∼35 events/h) that were synchronous with reoxygenations and accompanied by arousals. The authors found that CIH administered in this manner for 14 days augmented hypoxic and hypercpanic ventilatory responses, increased hematocrit and hemoglobin concentration, and raised daytime blood pressure.
This new model produces clinically relevant fluctuations in SaO2. Nevertheless, as the authors point out, there are several ways (e.g., poikilocapnia and no negative intrathoracic pressure development) in which their model does not mimic sleep apnea. We would add that the cardiovascular and respiratory consequences of this CIH paradigm also differ in several ways from the known effects of sleep apnea. For example, the percentage of sleep time at <90% SaO2 (∼80%) greatly exceeded that of most patients with sleep apnea, and the exposure caused increases in hematocrit and hemoglobin concentration and increases in the slopes of the ventilatory response to hypoxia and hypercapnia. The latter findings are not consistently seen in patients with sleep apnea (3, 16, 22). In these aspects, the model of Tamisier et al. more closely resembles the hypoxia of high-altitude exposure. Nevertheless, this innovative model can no doubt be refined to bring the pattern of intermittent hypoxia more in line with the disease it is intended to mimic. The model developed by Tamisier and colleagues represents an important step forward in the quest for methods that are realistic yet applicable to nonapneic human subjects who are free from cardiovascular, respiratory, or metabolic disease.
The report by Lee and colleagues (10), also in this issue of the Journal of Applied Physiology, reports on blood gas changes produced by an established mouse model of sleep apnea. Through rapid sampling of arterial blood in conscious mice, the authors defined the deoxygenation-reoxygenation profile associated with a CIH paradigm designed to resemble moderate to severe sleep apnea (60 events/h). In their model, FiO2 (measured at the nose) was reduced to 5–6% over 30 s followed by rapid reoxygenation to room air levels within the succeeding 30 s. This paradigm resulted in clinically relevant cyclical decreases in arterial Po2 (to 47 ± 2 mmHg) and SaO2 (to 85 ± 2%). The authors have previously demonstrated that most desaturations caused by this CIH paradigm are accompanied by arousals from sleep; however, this model does not mimic the fluctuations in intrathoracic pressure and arterial Pco2 that occur during episodes of sleep apnea. Nevertheless, this protocol, with its carefully quantified arterial oxygenation profiles, represents an important advance in the development of animal models of sleep apnea.
In the past, a rather wide variety of animal models has been used to study the cardiovascular consequences of CIH (1, 2, 4, 5, 7, 9, 11, 12, 19, 21) (see Table 1). The groundbreaking study of Fletcher and colleagues (4) that first demonstrated the hypertensive effect of CIH employed exposures that were severe with regard to the frequency of events (120 events/h) and level of FiO2 (2–5%). The episodes of hypoxemia in these experiments were accompanied by hypocapnia, not hypercapnia, and there were no large fluctuations in intrathoracic pressure secondary to airway obstruction. In what must be considered an experimental tour de force, Brooks and colleagues (2) developed a dog model that mimics all aspects of human sleep apnea. Using a computer-controlled tracheal occluder that was activated by EEG evidence of sleep and deactivated by arousal, they demonstrated substantial increases in daytime arterial pressure caused by the equivalent of “severe” sleep apnea (50–60 events/h for 14–16 h/day). This group of investigators also showed that arousals triggered by auditory stimuli, which were delivered at the same frequency and which caused comparable acute pressor responses as the tracheal occlusions, did not affect daytime blood pressure. More recently, CIH protocols that model “mild” or “moderate” sleep apnea in terms of desaturation frequencies have also produced increases in normoxic blood pressure (1, 5, 7, 12, 21). In most cases, the depth of desaturation produced by these protocols is not known; however, when data from these separate laboratories are viewed together, there seems to be a rough dose-response relationship between the frequency of CIH events and magnitude of blood pressure rise. The greatest increases in blood pressure were observed in studies in which hypocapnia was prevented by CO2 supplementation (1, 7, 21).
The validity of such a comparison is, of course, questionable. Each laboratory applied its own unique CIH paradigm; the frequency of events was only one of many differences in experimental conditions. These discrepancies make generalizations difficult, and, in our opinion, they hinder efforts to understand the cardiovascular, respiratory, metabolic, and cognitive effects of sleep apnea. To maximize progress in this field of inquiry, we believe that investigators should make a concerted effort to standardize their methods and employ the most realistic models available. Such an effort would require knowledge of the deoxygenation-reoxygenation profile, the arousal status, and changes in Pco2 associated with each protocol. Hypercapnia and abrupt changes in sleep state are likely to be important accompaniments of intermittent hypoxia in models of sleep apnea, because they would be expected to augment the amount of sympathetic activation evoked by each event (6, 13, 14). In contrast, negative intrathoracic pressure, because of its sympathoinhibitory effect (8, 15), is probably not critical for modeling the prohypertensive effects of sleep apnea. Second, humans and animals of both genders must be studied, given that there is some evidence that males and females respond differently to CIH (5). Finally, the cardiovascular consequences of CIH paradigms that model “mild” and “moderate” sleep-disordered breathing in terms of event frequency and depth of desaturations must be determined. Mild to moderate sleep-disordered breathing, consisting mainly of hypopneas with modest desaturations, is highly prevalent in the general population (23). Two prospective correlational studies (17, 18) have reached disparate conclusions about the effect of mild to moderate sleep-disordered breathing on the development of hypertension, indicating that cardiovascular risks associated with this amount of sleep-disordered breathing are as yet uncertain. Experimental evidence from animal and human models is required to elucidate such risks and inform clinical decisions about whom to treat.
- Copyright © 2009 the American Physiological Society