Sleep Spindles May Phasically Increase the Tonus Of the Upper Airway Dilator Muscles


John Zimmerman, Ph.D., DABSM, RPSGT; Madhukar Kaloji, M.D. FCCP; Michael Buxton, Ed.D.; Susan Jiannine, Rabindra Mohabir; Mark Hopkins, RPSGT and Laurel Kulm, RPSGT


We propose that the upper airway dilator muscles, especially the genioglossus muscle, receive tonic stimulation during wakefulness when alpha or beta activity is present in the EEG.  It has long been known that the upper airway tonic dilator muscle activity decreases during sleep onset.  However, we hypothesize that some phasic stimulation of the upper airway muscles does occur during sleep and that this phasic stimulation occurs during the time when sleep spindle are present in the EEG tracings.  We note the closeness and partial overlapping of the frequency ranges between alpha activity (8-13 Hz) and sleep spindles (11-16 Hz).  We also note that past experiments have shown that biofeedback or neurofeedback training of a wakefulness brainwave called the sensorimotor rhythm increases the number of sleep spindles in the sleep EEG.  Extrapolating from this data we propose that sensorimotor rhythm neurofeedback training during the day may reduce the severity of upper airway obstructions (apneas) or partial obstructions (hypopneas) occurring during the sleep period.


Sleep spindles are phasic, short bursts of brain wave activity that used to be defined in the Rechtschaffen and Kales Manual1 as 12 to 14 Hz bursts of activity lasting at least ½ second in duration and typically about a second or two.  A new 2007 definition of a sleep spindle was defined as a slightly broader range of frequencies in The AASM Manual for the Scoring of Sleep and Associated Events Rules, Terminology and Technical Specifications.  The new sleep spindle definition now used states that a sleep spindle is “A train of distinct waves with frequency 11-16 Hz (most commonly 12-14 Hz) with a duration of >0.5 seconds, usually maximal in amplitude using central derivations.” 2 Both sleep spindles and K-complexes are distinct brain waves that define the presence of Stage N2 sleep.  They are both present during stage N3 sleep as well but are usually absent during stage N1 sleep and REM sleep.


Surprisingly, no one has yet noticed the relationship between the frequent occurrence of sleep spindles in the EEG and the typical, usual absence (with two exceptions noted below) of obstructive sleep apnea/hypopnea in the respiratory channels of a polysomnogram (PSG).  More importantly, to the best of our knowledge, with the exception of only one paper no one has noticed that when sleep spindles are absent or sparse in the sleep EEG that obstructive sleep apnea/hypopnea is present.  One 1997 paper by Sembrano, Barthien, Wallace, and Lamm describes the results of a polysomnogram on a 23-year-old woman with mitochondrial encephalomyopathy with neurogenic muscle weakness, ataxia, and retinitis pigmentosa.  They stated, “An overnight polysomnogram (PSG) showed apnea, EEG slowing, and a paucity of sleep spindles” 3 but did not generalize their findings beyond this specific case.

We are presenting what we call the sleep spindle upper airway phasic EMG stimulation hypothesis.  We specifically propose that sleep spindles phasically increase the tonus of the dilator muscles of the upper airway.  As a consequence we propose that the more frequently sleep spindles occur, the more often the upper airway dilator muscles are phasically stimulated to increase their tone.  The more often the upper airway is phasically stimulated during the sleep spindle duration, the less likely the person is to have upper airway near-total collapse causing obstructive sleep apnea or upper airway partial collapse causing obstructive sleep hypopnea.  Two exceptions to this rule for obstructive sleep apnea/hypopnea and one exception regarding central sleep apnea/hypopnea will be noted later.



The elements of empirical evidence indirectly supporting this sleep spindle hypothesis are as follows:

1)      Obstructive sleep apnea/hypopneas and their resulting oxygen desaturations most often occur during transitions from wakefulness to Stage N1 sleep, when sleep spindles are absent.

2)      Obstructive sleep apnea/hypopnea often worsens during periods of REM sleep when again sleep spindles are largely absent and the muscle tonus of the upper airway decreases even further owing to the non-reciprocal tonic motor inhibition of muscle tone throughout the body.  In fact the sudden disappearance of both sleep spindles and K-complexes in the EEG is an indication that REM sleep is about to begin or has in fact begun.  When actual rapid eye movements occur in the presence of a slightly lower amplitude EEG signal (lower than that seen during stage N1 sleep), a relatively more uniform amplitude EEG pattern (more uniform less variable EEG amplitude than that present during stage N1 sleep EEG), a low tonic chin electromyogram (EMG) signal, then REM sleep is unequivocally present.

3)      Within the same night when a patient with obstructive sleep apnea/hypopnea has alternating periods of normal breathing and obstructive sleep apnea/hypopnea it is clear that during normal breathing and while oxygen saturation levels are the highest and the most stable that sleep spindles occur frequently in the EEG.  Conversely, during the same night when obstructive sleep apnea/hypopneas and their resulting oxygen desaturations are present, then during that time period sleep spindles are usually absent or sparse.


This is a hypothesis that is clearly open to experimental testing by simultaneously recording the EMG of some of the muscles of the upper airway; say the genioglossus muscle, while recording a regular PSG with the sleep EEG and breathing patterns.  If this hypothesis is correct, there should be a recordable phasic increase in the genioglossus EMG each time a sleep spindle occurs.  Furthermore, we would predict a tonic increase in genioglossus EMG during periods of time of alpha (8-13 Hz) or beta activity in the EEG as well, i.e. while the patient is awake.


The alpha rhythm and sleep spindles share some similarities.  The alpha rhythm has an AASM-defined frequency of 8-13 Hz, whereas the sleep spindle frequency range is 11-16 Hz.2 These frequency ranges somewhat overlap in the 11 to 13 Hz range.  Because of the overlap in frequencies sometimes long duration sleep spindles are difficult to distinguish from alpha activity if topographic data is not available.  Alpha activity is concentrated in the occipital and parietal areas.  Sleep spindles are most well developed over the central areas of the brain such as at the C3 and C4 electrode locations.  Both the alpha rhythm and sleep spindles depend upon recursive thalamocortical feedback networks4,5.  We hypothesize that alpha activity tonically stimulates the upper airway dilator muscles over a period of seconds, to minutes to as long as hours in duration during EEG arousals and during wakefulness.  Sleep spindles stimulate the upper airway dilator muscles in a shorter durations (0.5 to 3 second time intervals) phasic manner.


Thus this hypothesis explains why the alpha activity of an EEG arousal following an apnea or hypopnea is usually associated with increased amplitude air flow signals as it predicts higher tonic airway dilator muscle activity during alpha or beta activity.  This alpha-induced tonic stimulation of the upper airway dilator muscles thus eliminates the upper airway near total obstruction (apnea) or partial upper airway collapse (hypopnea).  Conversely when both tonic alpha or beta activity is absent and phasic sleep spindle activity is also absent as is the case during the EEG slowing phase of sleep onset during the wake-to-stage N1 sleep transition, this allows upper airway collapse to occur.  Thus we are postulating that in the presence of either a stage N1 EEG or a stage REM EEG pattern, when sleep spindles are absent, and when the EEG consists primarily of only low amplitude, mixed frequency EEG patterns that this allows obstructive sleep apneas or hypopneas to occur.


Normally, at the beginning of the night, perhaps the reason why stage N1 sleep is so short is that it is to the advantage of a sleeping person to quickly pass through the vulnerable stage N1 sleep stage, when sleep spindles are absent.  It is best, from a respiratory standpoint, to enter stage N2 sleep quickly so that the airway-dilating attributes of sleep spindles can phasically stimulate the muscles of the upper airway to prevent the occurrence of obstructive sleep apnea/hypopnea.  In fact someone suggested at an APSS conference several years ago that if a patient could get out of stage N1 sleep and quickly into stage N2 sleep, using a sedative hypnotic, that this might help to reduce the severity of or perhaps even alleviate obstructive sleep apnea/hypopnea6.


Two exceptions to the protective effects of sleep spindles upon obstructive sleep apnea/hypopnea have been noted.  Specifically, obstructive sleep apnea/hypopnea may occur even while sleep spindles are present under two conditions: 1) When the apneas or hypopneas are extremely long and 2) when the apneas and hypopneas are extremely frequent as is the case with patients who have very severe obstructive sleep apnea/hypopnea with an apnea/hypopnea indexes of 30 or more apneas or hypopneas per hour of sleep.


The first exception occurs when a hypopnea (or rarely an apnea) is extremely prolonged that is over 60 seconds in duration and sometimes an event that is three minutes or longer measured from start to finish.  Usually such super-long duration events often occur during REM sleep but if they happen to occur during stage 2 sleep, then the patient may be asleep long enough to shift the brain wave patterns from a low voltage mixed frequency EEG pattern (Stage N1 sleep) into spindle sleep (Stage N2 sleep).  The same reasoning explains why a patient on CPAP therapy, who had not had many sleep spindles while untreated and while breathing just plain room air does have frequent sleep spindles while being treated with CPAP.  It is simply because the patient is able to sleep long enough to shift the brain wave patterns from stage N1, without sleep spindles, to stage N2, with sleep spindles.


The other exception occurs when the apneas or hypopneas occur extremely frequently once every other minute or so.  This would represent an apnea/hypopnea index of 30 apneas or hypopneas per hour of sleep when an event occurs about every other minute while asleep.  When a respiratory disturbance begins with about every fourth 30-second epoch, then the EEG arousals become so frequent and sleep continuity becomes so disrupted with literally hundreds of EEG arousals per night that a significant amount of sleep deprivation occurs.  This naturally builds up a great deal of sleep pressure.  This may cause the patient to enter Stage N2 sleep much more quickly than a person with a more mild degree of sleep-disordered breathing, less sleep disruption, and a more typical amount of sleep pressure.  Such severe OSAH patients, with an AHI of 30 or more, sometimes do show sleep spindles during apneas or hypopneas simply because they are so extremely sleepy.


Neither one of these exceptions, though, that is extremely prolonged events or extremely frequent events should distract from the hypothesis that sleep spindles may under more usually encountered circumstances phasically stimulate the muscles of the upper airway and usually (but not always capable) prevent the occurrence of sleep apnea or sleep hypopnea (except when the events are extremely prolonged or frequent).  In such extreme cases of sleep disordered breathing tonic stimulation of the muscles of the upper airway, such as occurs when alpha or beta activity is present in the EEG, may be necessary to keep the upper airway patent rather than just phasic stimulation as afforded by sleep spindles.


Another major exception to this theory occurs with central, mixed, or complex sleep apnea.  Sleep spindles are frequently present during periods of central sleep apnea/hypopnea such as in patients with idiopathic central sleep apnea or complex sleep apnea brought about by initial therapy with continuous positive airway pressure (CPAP).  This is because the pathophysiological mechanisms underlying the development of central sleep apnea/hypopnea are different than for obstructive sleep apnea/hypopnea.  In central apnea there appears to be an increased hysterisis of the feedback control loop between the pCO2 set point of the waking state, say 45 mm Hg and the higher pCO2 set point of the sleeping stage, say 49 mm Hg.


At the transition from wakefulness to sleep minute ventilation decreases, causing the PCO2 level to increase.  Thus at sleep onset the pCO2 level may be at the waking value stimulatory of respiration say at 45 mm Hg but if it remains at 45 mm Hg at sleep onset and is below the sleep pCO2 set point to breathe, say at 49 mm Hg, then a central apnea/hypopnea may develop because the hypercapneic drive is already satisfied.  The patient will simply not breathe until the pCO2 level builds up to the sleep set point value of, say 49 mm Hg, and then breathing will resume with no deep gasp or large breath.  Sleep spindles may still readily be observed during repetitive periods of central sleep apnea/hypopnea and may still be phasically stimulating the muscles of the upper airway but to no avail since the hypercapneic drive has been blunted by the transition from wakefulness to stage N1 sleep and the changing pCO2 set point to breathe.


Thomas, Daly, and Weiss provided evidence that blunting of the hypercapneic drive leads to central sleep apnea by adding tiny amounts of CO2 to a CPAP mask to a patient experiencing complex sleep apnea and repetitive central sleep apnea/hypopnea episodes7.  Upon adding just 0.5 to 1.5% of carbon dioxide to the CPAP circuit thus increasing TcCO2 from 46.3 to 49.3 mm Hg, the central apneas/hypopneas disappeared.


We further hypothesize that complex sleep apnea may be an iatrogenic problem caused by inadvertent lowering of the pCO2 via dozens of liters per minute of fresh air blowing off the CO2 that otherwise might be inhaled by a sleeping patient with their nose partially sunk into a pillow or covered with a sheet or blanket.  The added pCO2 resulting from re-breathing one’s exhaled carbon dioxide with the patient’s nose adjacent to a pillow could stimulate the hypercapneic drive just enough to prevent the occurrence of central apneas or central hypopneas (shallow, in-phase breathing with no snoring).  In addition, during initial CPAP therapy if the average oxygen saturation level increases say from somewhere in the 80s on room air to somewhere in the 90s while on CPAP this would blunt the hypoxic drive as well thus also leading to central apneas/hypopneas.  Lastly, inflating the lungs more than usual with 30 to 60 or more liters per minute of pressurized air entering the upper airway is likely to stimulate the muscle spindle receptors of the intercostal muscles thus stimulating the Hering-Breuer reflex, which is also likely to cause a central apnea/hypopnea.


There is, in the biofeedback literature, a brain wave type called the sensorimotor rhythm or sensory-motor rhythm (SMR).  In humans, the SMR has a frequency of 9-13 Hz8.  This frequency is similar to the new definition of the sleep spindle frequency of 11-16 Hz.  Back in the early 1970s M.B. Sterman showed that SMR biofeedback training could reduce the occurrence of epilepsy and seizure activity9,10.  Given the similarities between the sleep spindle frequency (11-16 Hz) and the SMR frequency (9-13 Hz) it should be no surprise that biofeedback or neurofeedback training of the SMR during several days resulted in an increase in spindle-burst sleep (stage N2) at night.  Thus it appears that neurofeedback training in the waking state of the SMR rhythm at least somewhat generalized to the sleeping state thus increasing the number of sleep spindles at night11.  Given the proposed protective effect of sleep spindles through purported phasic stimulation of the muscles of the upper airway, such neurofeedback training to increase the amount of sensorimotor rhythm could possibly be a novel treatment for obstructive sleep apnea/hypopnea.


We would welcome any experimental verification of this sleep spindle-upper-airway EMG-stimulation hypothesis.  An animal model is one method to experimentally prove or disprove this hypothesis by artificially increasing or decreasing the number of sleep spindles, say with benzodiazepine class of drugs, and to then correlate the increase in benzodiazepine-induced sleep spindles with increased phasic stimulation of the muscle tone of the upper airway muscles.




  1. Rechtschaffen A, Kales A. A Manual of Standardized Terminology, Techniques and Scoring System for Sleep Stages of Human Subjects, 1968; US Department of Health, Education, and Welfare Public Health Service – NIH/NIND.
  2. Iber C, Ancoli-Israel S, Chesson A, and Quan SF for the American Academy of Sleep Medicine. The AASM Manual for the Scoring of Sleep and Associated Events Rules, Terminology and Technical Specifications, 2007; 1st ed: Westchester, Illinois: American Academy of Sleep Medicine.
  3. 3. Sembrano E, Barthlen, GM, Wallace S. and Lamm C. Polysomnographic findings in a patient with the mitochondrial encephalomyopathy NARP.  Neurology, 1997 Dec; 49(6): 1714-7.
  4. Lindsley, DB. Foci of activity of the alpha rhythm in the human electro-encephalogram. J Exp Psych, 1938; Vol 23(2):159-71.
  5. Velasco, M and Lindsley, DB. Role of orbital cortex in regulation of thalamocortical electrical activity. Science, 1965; 149(690):1375-7.
  6. Presentation of someone at an APSS conference suggesting using a sedative-hypnotic to help get a patient out of stage N1 sleep quickly and into stage N2 sleep as a way of treating obstructive sleep apnea/hypopnea.
  7. Thomas, R.J., Daly, Robert W. and Weiss, J. W. Low-concentration carbon dioxide is an effective adjunct to positive airway pressure in the treatment of refractory mixed central and obstructive sleep-disordered breathing.  Sleep, 2005; Vol 28, No. 1: 69-77.
  8. Kropotov, J. Quantitative EEG, event-related potentials and neurotherapy, 2008; Academic Press, 542 pages.
  9. Sterman, M.B. and Friar, L. Suppression of seizures in epileptic following sensorimotor EEG feedback training.  Electroenceph. Clin. Neurophysiol, 1972; 33:89-95.
  10. Sterman, M.B., Macdonald, L.R., and Stone, R.K. Biofeedback training of the sensorimotor EEG rhythm in man: Effects on epilepsy, Epilepsia, 1974; 15:395-417.
  11. Sterman, M.B., Howe, R.D. and Macdonald, L.R. Facilitation of spindle-burst sleep by conditioning of electroencephalographic activity while awake.  Science, 1970; 167:1146-1148.


Please address all correspondence to Madhukar Kaloji, M.D., F.C.C.P. and John Zimmerman, Ph.D. at 1020 Independence Blvd. Suite 205, Virginia Beach, Virginia 23455.  May 11, 2011

Share and Enjoy:
  • Print
  • Digg
  • Sphinn
  • Facebook
  • Mixx
  • Google Bookmarks
  • Blogplay

Related posts:

  1. Comparative Effects of Two Oral Appliances on Upper Airway Structure in Obstructive Sleep Apnea
  2. Practice Parameters for the Surgical Modifications of the Upper Airway for Obstructive Sleep Apnea in Adults
  3. Sleep bruxism related to obstructive sleep apnea: the effect of continuous positive airway pressure.

Leave a Reply

Your email address will not be published. Required fields are marked *