SUMMARY

In OSA, daytime symptoms (sleepiness, impaired thinking, altered mood) have been attributed to both nighttime hypoxemia (see Glossary) and to sleep fragmentation by cortical or EEG arousals (see Glossary). The strong connection of the two in OSA makes it hard to gauge the contribution of each. In normals, external stimuli during sleep for a single night can cause similar symptoms. Furthermore, in mild sleep apnea, simple snoring, and upper airway resistance syndrome similar symptoms occur (and improve with CPAP); in these cases, there are sleep disruptive EEG arousals without hypoxemia. However, in OSA itself, sleep fragmentation by arousals accounts for only about 16% of the variance in daytime sleepiness. In fact, many symptomatic OSA patients have no more frequent arousals than symptom-free normal subjects. Not all respiratory events result in arousals; some respiratory events may be followed only by slight blood pressure or heart rate increases to signify a physiological response. The authors hypothesized that such sleep disturbances, indicated by blood pressure or heart rate changes without meeting criteria for EEG arousals, might be enough to impair daytime function in normal subjects. The study group consisted of 12 student volunteers, 7 men and 5 women, averaging 25 years old (SD=+/-6 yrs). All were nonobese, with an average Body Mass Index (see Glossary) of 22 (SD=+/-2). They all had normal Epworth Sleepiness Scores (see Glossary) averaging 5, ranging from 2 to 8. A sleep-wake questionnaire was used to screen out those with likely sleep disorders. All subjects underwent two successive weeks of 2 nights and 1 day in the laboratory, the first night to “acclimate” to the lab, the second night for experimental sleep disturbance by auditory stimuli (tones). Half the subjects experienced these tones the second night of the first week, the other half the second night of the second week. Throughout the night of auditory stimulation, the volume and duration of the tones were adjusted, generally upward to offset “acclimatization” to the stimuli. At the same time, tones were adjusted in an effort to produce no EEG changes sufficient to be rated as an “arousal,“ (see Glossary) but an “autonomic” (see Glossary) response manifested by at least a 4 mm Hg increase in systolic blood pressure or at least a 4 beat per minute increase in heart rate. This autonomic response had to occur within 15 seconds of a tone and persist for at least 3 heartbeats. EEG changes of less than 3 seconds duration were not counted as arousals, even if they included sleep spindles (see Glossary) or K-complexes (see Glossary), which were often observed in response to these tones. Instead, this study defined an EEG or cortical “arousal” as a return to alpha or theta frequency of the EEG for at least 3 seconds, regardless of sleep stage. Note that this may not exactly correspond to operational definitions of arousal used in other laboratories. If a given duration and intensity of tone stimulus produced this type of autonomic but non-cortical “arousal,” it was repeated one minute later. If, on the other hand, the tone produced a cortical arousal, the experimenter waited until sleep was reestablished, then waited another minute before applying the tone again. If the tone evoked no response at all, neither autonomic nor cortical, after 30 seconds the next tone was emitted at higher volume or for longer duration. The intent was to establish autonomic, non-cortical “arousals“ at 1 minute intervals throughout sleep. The day after each second night, subects underwent testing of daytime sleepiness, mental functioning, and mood. Daytime sleepiness was assessed with the Multiple Sleep Latency Test (see Glossary) and the Maintenance of Wakefulness Test (see Glossary), each consisting of 4 naps at 2-hour intervals terminated after one epoch of Stage 1 sleep to avoid cumulative recovery of any sleep lost the night before. Subjective Sleepiness was assessed with the Stanford Sleepiness Scale (see Glossary) administered at 7:00 a.m. and prior to each MSLT nap. Mood was assessed with an adjective checklist at the same time intervals. Prior to the first nap of the day, subjects underwent performance tests drawn from the Wechsler Adult Intelligence Scale-R (digit-symbol substitution and block design), as well as trailmaking tests, the steer clear test, and a paced auditory serial addition test. These tests yielded a large number of outcome measures which were condensed by a statistical procedure called principal-components analysis into a one or a few combined scales showing major, independent effects from the experimental manipulation (i.e., sleep disturbance by tones vs. undisturbed sleep). The disturbed and undisturbed nights did not differ on total sleep time, amounts of wakefulness, or amounts of stage 1, 2, or REM sleep. However, the tone-disturbed night showed significantly less slow-wave (stages 3 and 4) sleep: 19.5% (SD=+/-4.3) versus 23.9% (SD=+/-5.0) on the undisturbed night. During the disturbed night, an average of 253 tones (SD=+/-23) were presented to each subject, of which 44% produced EEG arousals, amounting to an average frequency of tone-induced cortical arousals of 7.8 (SD=+/-2.2) per hour of sleep. However, on the undisturbed night there was virtually the same frequency of total cortical arousals per hour of sleep as on the fragmented night (22-26/hr vs. 24-27/hr) and the same frequency of actual awakenings (3.7 vs. 3.8). EEG alpha rhythm, one sign of arousal, increased more after tones which induced visible EEG arousals than after tones which did not induce such arousals. The day after “nonvisible“ auditory sleep disturbance, there were significantly shorter average latencies to sleep onset on the MSLT (6.2 minutes SD=+/-2.1, vs. 8.0 mins SD=+/-3.1) after the undisturbed night. Likewise, on the MWT the latency of sleep onset was shorter after sleep disturbance (25.7 SD=+/-9.7) than after undisturbed sleep (29.0 SD=+/-10.0). There was no significant difference in subjective daytimes sleepiness measured by the Stanford scale after disturbed versus undisturbed sleep. Mood was lower in the early morning after disturbed than after undisturbed sleep. A single main cognitive factor that came out of the principal components analysis did not show any greater impairment after disturbed than after undisturbed sleep. The authors took their results to mean that “nonvisible“ sleep fragmentation by tones affects sleep architecture (by reducing the proportion of slow-wave sleep), increases daytime sleepiness (as measured by the MSLT and MWT), and lowers mood on awakening. Although there was no increase in “visible” cortical arousals on the night of disturbed versus undisturbed sleep, the authors did note small but significant changes in EEG alpha rhythm and total EEG power in response to tones that didn’t cause visible EEG arousals. They noted that other research has shown the magnitude of the adverse change in sleepiness induced in normal subjects by tones that produce visible arousals is greater than the magnitude of therapeutic change resulting from CPAP in OSA patients, but that the change induced in normals by nonvisible arousals is less than the change induced by CPAP. In other words, the magnitude of effect from this particular type of experimental sleep disturbance is expected to be less than that seen in clinical situations. Perhaps this is why they failed to observe cognitive impairment in their sleep-disturbed subjects. The authors suggested that a more sophisticaled analysis of the EEG or a “nonvisible“ marker of arousal should be included in routine sleep studies to assess the real extent of sleep fragmentation in sleep disorder patients.

COMMENTS

This study is complex, not so much in its basic design, simply a crossover of all subjects from tone-disturbed sleep one week to undisturbed sleep the other week, with assessment of sleepiness, mood, and mental function the next day. Rather, the complexity has to do with the definition of EEG or “cortical” arousal and the converse definition of “non-visible” arousals as lacking the defining features of the visible (that is, scorable) arousals. It is important to realize from the outset that “arousals” in sleep medicine are nothing like what we mean by “awakenings,” which are much less common both in patients and in normals. Arousals generally signify a transition from one stage of sleep to another, without return to consciousness. Their exact definition is open to some differences of opinion and practice, to say nothing of depending on the skill and trained agreement of individual raters at a particular laboratory. To put it most personally, I have had the experience of the rater at one reputable center failing to identify PLMS-related arousals which were obvious to me when pointed out by another rater at another reputable lab, and also obvious to the director of the original lab once they were pointed out to her. The hard copy of my polysomnogram had been given to me as exemplifying “normal sleep” when actually it illustrated substantial sleep fragmentation by PLMS. When these authors refer to visible arousals, they mean a fairly gross level of arousal, even though they don’t include the EMG measures ordinarily factored in to judgements about arousals. For example, they do not consider K complexes or sleep spindles occuring in slow-wave sleep to represent arousals if they are very brief. Their conclusions reflect a need for more sensitive and subtle measures to define EEG arousals and one could suggest the same about their own procedures. However, one odd result that seems to conflict with this view is that the frequency of arousals, both on disturbed and on undisturbed nights, seems unusually high, over twenty, though with great variation. One does not understand how 44% of arousals on disturbed nights could be associated with tones, without resulting in any increase of arousals on disturbed versus undisturbed nights. Perhaps the authors take this to mean that the tones actually had only a coincidental relationship to the arousals. One then wonders how to account for the frequent arousals. One point of importance that comes out of this line of research is that factors quite different from respiratory events, such as auditory stimuli, can yield sleep disruption comparable to sleep apnea, without the return to consciousness that would leave the person aware of the cause of sleep disturbance, or even aware of the existence of sleep disturbance. I think it has often been assumed that when such external stimuli interfered with sleep continuity, the individual would be aware of that process, i.e., the effects of a restless bed partner or a pet kept in the bedroom at night, a noisy neighbor upstairs, or a noisy airport or turnpike nearby. Apparently, no such awareness need exist. Therefore, the sleep clinician interviewing the patient for possible sources of sleep disruption cannot rely on a simply denial of noise, movement, temperature changes, odors, lights, or even internal bodily stimuli like gas, cramps, or esophageal regurgitation, to judge whether sleep at home might be disrupted by such factors. Furthermore, many such factors would not exist in the laboratory. In my discussions of several papers, I have reiterated the point that sleep continuity vs. fragmentation should be viewed as a multifactorial problem, where resolution of one major component such as the respiratory events with CPAP may still leave sleep disturbed by other stimuli, such as the leakage of air from the CPAP mask itself, the periodic leg movements that sometimes appear as apneic events are controlled, or other disturbing internal and external stimuli quite unrelated to OSA or CPAP. These other stimuli may be sufficient to cause clinically significant daytime symptoms by themselves; or they may require summation with other, “minor” factors to yield similar results.





This page hosted by Get your own Free Home Page