Cortico-motoneurone excitability in patients with obstructive sleep apnoea


Prof. Roberto Cantello, Dipartimento di Scienze Mediche, Via Solaroli 17, 28100 Novara, Italy. Tel.: +393487223948;
fax: +3903213733298; e-mail:


A disordered neuromotor control of pharynx muscles may play a role in the genesis of obstructive sleep apnoea syndrome (OSAS). This raises the possibility of a dysfunction of projections descending from the cortex to segmental nuclei. With single pulse transcranial magnetic stimulation (TMS) we studied the physiology of the corticospinal projection to hand muscles in seven OSAS patients. At first, we compared them with nine age- and sex-matched normal controls in the wake state. The only abnormality was a lengthening of the central silent period (P < 0.001). This supports a steady imbalance of motor cortical interneurone activities towards a state of enhanced inhibition. Then we looked at changes of the motor-evoked potential (MEP) size and latency, according to whether patients were awake, or in a non-rapid eye movement (REM) 2 sleep stage, or during a typical apnoea. During non-REM 2 sleep, the average MEP amplitude was significantly (P < 0.05) smaller than in the awake state. The MEP latency was, in turn, significantly longer (P < 0.05). During apnoeas, the MEP size decreased, and the latency increased further (P < 0.05), indicating an extra depression of the cortico-motoneuronal activity. All TMS changes were detected outside the pharyngeal district, suggesting a widespread dysfunction of the cortico-motoneuronal system in the OSAS, which is more evident during apnoeas.


Obstructive sleep apnoea syndrome (OSAS) consists of repetitive upper airway obstructions during sleep. Most commonly, it includes excessive daytime somnolence and cognitive deficits, and is prone to cardiovascular complications (Cetel and Guilleminault 1994). The OSAS pathophysiology is somehow obscure, although there are some clear risk factors such as obesity (White 1995). In principle, the contractile activity of the pharyngeal dilator muscles is not defective in OSAS. However it is particularly vulnerable to fatigue. Moreover, it shows a net drop during sleep (McNicholas 2003). A disordered motor control of these muscles was proposed as a pathogenic OSAS factor (Horner 1996). Physiological abnormalities may be segmental or suprasegmental (Horner 2000). The latter may result from a malfunction of the cortico-motoneuronal system and/or of the brainstem centres projecting to the bulbar motor nuclei. Breathing-linked alterations of cortical functioning were described previously in OSAS (Gora et al. 2002). In a preliminary report, Cegla and Fröde (1993) used transcranial magnetic stimulation (TMS) to explore the cortico-motoneurone physiology in this syndrome. The latencies of the motor-evoked potentials (MEPs) in hand muscles were longer than normal. This suggested some widespread defect in the conductivity or excitability of the system. We wanted to further explore such a hypothesis in a group of OSAS patients. At first, we contrasted their TMS findings while awake with those of normal controls matched for sex and age. Secondly, we looked at within-patient MEP changes during ‘normal’ slow sleep and during apnoeas.



We studied seven consenting patients with essential obesity. They showed chronic hypoxia/hypercapnia. All met the criteria for the obesity-hypoventilation syndrome in which OSAS is the rule (Kessler et al. 2001) (Table 1). Indeed, their clinical-polysomnographic findings were typical of a severe OSAS (Strohl and Redline, 1996). Excessive day-time sleepiness (average score on the Epworth Sleepiness Scale = 17.3 ± 2.3) and nocturnal snoring were always reported. Patients were not taking CNS active drugs or had a positive airway pressure treatment previously. They had no chronic obstructive pulmonary disease. Computed tomography/magnetic resonance imaging scans of the brain were normal.

Table 1.  Main features of OSAS patients while awake and controls
FeatureOSAS patientsControlsP-value
Age32.7 years (SD 12.7) 36.4 years (SD 10.3) 
SexFour men, three womenFive men, four women 
Body mass index65 kg m−2 (SD 6.7)  
SaO285% (SD 4.7)  
PaO266.5 mmHg (SD 6.8)  
PaCO250 mmHg (SD 5.8)  
Average relaxed motor threshold (% of stimulator output)52.6% (SD 3.6)47.5% (SD 7.9)NS
Average MEP size (mV)4.6 (SD 2.1)5.1 (SD 2.4)NS
Average MEP latency (ms)21.2 (SD 0.6)20.8 (SD 0.8)NS
Average duration of the central silent period (ms)232.1 (SD 15.8)179.5 (SD 17.5)<0.001

Transcranial magnetic stimulation

The TMS studies was conducted along the standard criteria published by the International Federation of Clinical Neurophysiology (Rothwell et al. 1999). Briefly, we used a Magstim 200 stimulator (Magstim Co., Whitland, Dyfed, UK) with a round flat coil centred horizontally at the vertex. For optimal coil placing, the vertex was clearly marked on the electroencephalogram (EEG)-recording cap. This also allowed precise coil repositioning if involuntary skull displacements occurred. An anticlockwise current (coil viewed from above) excited the left motor cortex. MEPs were recorded from the right first dorsal interosseus (FDI) muscle. Signals were stored and analysed by a Neuroscan machine (Neuroscan laboratory, Sterling, VA, USA).

Wake experiment

Patients lay supine in a comfortable chair with their eyes closed, in a condition of maximum muscle relaxation. We wanted to evaluate a series of TMS variables that admittedly inform on the cortico-motoneuronal excitability (Abbruzzese and Trompetto 2002). Thus we measured the relaxed motor threshold (rMT) on a standardized basis (Rothwell et al. 1999). We also averaged the peak-to-peak MEP size, and the MEP latency in the resting FDI following 15 TMS shocks, whose intensity was 1.5 × rMT. Identical stimuli were then used to determine the duration of the central silent period (cSP, average of 15 trials). In most cSP studies, the background electromyogram (EMG) was between 10 and 20% of the maximum voluntary contraction (MVC) (Reid et al. 2002). We used 10% MVC, a level that we quantified as previously reported (Civardi et al. 2001). Subjects were instructed to sustain their background contraction even after having perceived the magnetic shock. The cSP duration went from the stimulus artefact to the consistent reappearing of EMG activity. Data obtained for rMT, MEP size, MEP latency and cSP were compared with those of controls by means of a univariate anova model (significance: P < 0.05).

Sleep experiment

For the sleep study, afternoon naps were sufficient as day-time sleepiness was quite prominent, and patients had a high number of apnoeas during naps (mean apnoea/hypopnoea index = 64 ± 13.6/h). Somnographic recordings included: EEG (C4-A2, O2-A2, C3-A1, O1-A1), electrocardiogram (ECG), electro-oculogram, EMG (chin and FDI muscles) and SaO2. Airflow in the nasal and oral cavities was monitored with thermistors. Thoracoabdominal respiratory motion was monitored with a strain gauge. During the sleep experiment, we periodically checked the electrode impedance, to ensure optimum recording conditions in spite of sweating or other disturbing variables. We then determined the MEP amplitude and latency in three patient states, which were later classified based also on the Rechtschaffen and Kales criteria: wake, stage 2 of non-rapid eye movement (REM) sleep, and apnoeas during stage 2. The order with which the three conditions were studied varied at random to avoid any sequence effect. We delivered 10 TMS stimuli whose intensity was 1.6 × rMT. During apnoeas, the TMS shock was given when the mean SaO2 was between 50 and 40%. The actual SaO2 level of each patient was recorded and grouped values were obtained. The peak-to-peak MEP amplitude and the MEP latency were computed, and grouped data were compared by means of a univariate anova model, with Bonferroni post hoc tests (significance: P < 0.05). All variables were expressed as mean ± SD. It was our purpose to record MEPs at rest. Thus, we analysed offline the prestimulus activity in the target muscle (at least 50 ms). If any amount of EMG was present, which seldom happened during the final period of apnoeas, we discarded the corresponding tracing. However, an accurate measurement of the relaxed threshold needs a much larger number of consecutive stimuli impinging on a muscle at complete rest (Rothwell et al. 1999). As it was sometimes difficult to fulfil these criteria during apnoeas, we did not consider threshold in this experiment.


In the awake condition, there was no significant difference in the rMT, the average peak-to-peak MEP size or the MEP latency between patients and controls. By contrast, the cSP duration was increased in the patients (P < 0.001) (Table 1).

As to the within-patient sleep study, the average group value of SaO2 during stage 2 of non-REM sleep was 86 ± 3.7%. In this condition, the average MEP size was significantly smaller (50%) than during the awake state (2.3 ± 0.97 mV versus 4.6 ± 2.1 mV; P < 0.05). In turn, the MEP latency was prolonged (22.4 ± 0.4 ms versus 21.1 ± 0.4 ms; P < 0.05) (Figs 1 and 2). During apnoeas, the average group value of SaO2 was 45 ± 4.2%. The average MEP size was 0.46 ± 0.15 mV, which corresponded to 10% of the awake value and to 20% of the non-REM 2 value. Such a reduction was significant (P < 0.05) when compared with both the above-mentioned conditions. The average MEP latency (24 ± 0.4 ms) showed a further prolongation during apnoeas, which was again significant (P < 0.05) when compared with both the awake and the non-REM 2 state (Figs 1 and 2).

Figure 1.

Typical examples of the MEP size and latency changes during relaxed wake (W), stage 2 of non-REM sleep (nR2), and an obstructive apnoea (A) in three OSAS patients. Tracings represent the overlap of 15 MEPs in wake, and 10 in non-REM2/apnoeas. Average SaO2 values (±SD) in the three different conditions are also shown.

Figure 2.

(a) Boxplots representing the range of the MEP amplitude values during relaxed wake, stage 2 of non-REM sleep, and an obstructive apnoea in the OSAS patient group. Apnoea values were significantly (*P < 0.05) reduced in comparison with both wake and non-REM2 sleep. (b) Boxplots representing the range of the MEP latency values during relaxed wake, stage 2 of non-REM sleep, and an obstructive apnoea in the OSAS patient group. Apnoea values were significantly (*P < 0.05) increased in comparison with both wake and non-REM2 sleep. Mean values are represented by bold horizontal line. Bars = ±SD.


Strong evidence supports the view that single-pulse TMS does provide information about the excitability of the cortico-motoneuronal system (Abbruzzese and Trompetto, 2002). As it is simpler and faster than other more complex TMS techniques, it can be used effectively during sleep (Grosse et al. 2002). The cSP is an interruption in the voluntary background activity that follows the MEP. The length of its late portion is meant to reflect complex motor cortical inhibitory phenomena (Cantello et al. 1992), possibly linked to the GABAergic tone (Boroojerdi, 2002). The silent period was found prolonged in many pathological conditions, such as for instance Huntington's chorea (Modugno et al. 2001), or in the motor cortex contralateral to area 4 foci in epileptic patients (Cincotta et al. 1998). Invariably, the interpretation was of an imbalance of cortical excitability towards a state of enhanced inhibition. We must then apply this view to the prolonged cSP of our OSAS patients while awake. The chronic hypoxia/hypercapnia could produce a shift towards inhibition in motor cortical circuits. Another factor might be the cortical excitability adjustments related to the neural basis of sleepiness per se. Chronic hypoxia has not been studied in terms of brain excitability. Acute anoxia in the animal is associated with a depression, and not an enhancement, of cortical inhibitory phenomena (Krnjevic et al. 1966). On the contrary, acute hyperoxia/hypocapnia is associated with an excess excitability in the human brain (Seyal et al. 1998). Episodic sleepiness after sleep deprivation did not significantly affect the silent period (Civardi et al. 2001). However it might well be that chronic effects of hypoxia or excessive daytime sleepiness may turn out quite different from the acute.

For a fixed stimulus intensity, the average MEP size is directly proportional to the number and density of cortical descending neurones ready to fire at that particular time, to the number of spinal α-motor neurones susceptible of being discharged by the descending volley, and to the synchronicity of both. Because of the intrinsic physiological properties of the corticospinal system, the MEP latency shows an inverse relation to these same factors (Rothwell et al. 1999). The MEP size has been widely used as an index of cortico-motoneuronal excitability, for instance in studies of Parkinson's disease (Cantello et al. 2002). In our patients there was an MEP size depression during non-REM 2 sleep and an increase of the MEP latency, which was similar to the findings of Grosse et al. (2002) in normal subjects. In fact, these authors described a reduction of the cortico-motoneuronal system excitability, which was maximum during REM sleep, and was more pronounced in stage 2 than in stage 4 of non-REM sleep. Such reduction may arise from enhanced spinal inhibition because of specific brainstem influences. However changes in the amount of cortical inhibitory processes were proposed as well (Khatami et al. 2001).

During apnoeas, the MEP size decreased, while its latency increased further. This represents an extra reduction of the cortico-motoneuronal excitability. At that time, it was unfeasible to test F- or H-waves for thoroughly assessing spinal excitability. Then, we cannot dissociate segmental (as well as brainstem) phenomena from cortical phenomena. Many factors might explain a cortical depression, such as, for instance, a reduction in brain electrical activity related to changes in cerebral tissue oxygenation and blood flow (Walsleben et al. 1993; Hayakawa et al. 1996), or an increase in intracranial pressure (Jennum and Boergesem 1989). Then, studies of normal subjects showed that fatigue produces a significant reduction of the MEP size through a ‘central’ mechanism (Di Lazzaro et al. 2003). Excess MEP size reduction (and latency increase) during apnoeas might then be one counterpart of the pathological muscle fatigue that is an important pathophysiological factor of OSAS (McNicholas 2003).

However, our experiments cannot specify if the TMS changes seen were a cause or an effect. The fact remains that, when awake, OSAS patients showed enhanced motor cortical inhibition. Apnoeas were associated with a further drop in the cortico-motoneuronal excitability, the precise level of which cannot be identified. We found such alterations in a neural district different from that subserving the pharynx. Unfortunately, MEP recordings from pharynx muscles are somewhat invasive. As such, they do not appear feasible in sleep. Moreover, the MEP size in this region is quite small, even while awake (Hamdy et al. 1996). The relation of the FDI EMG activity to breathing is obviously loose. More proximal muscles would seem a better candidate for TMS studies. However some of them, such as trapezious, show spontaneous EMG oscillations during sleep, which might distort our measurements (Westgaard et al. 2002). At the same time, stronger stimuli may be necessary for their excitation (Reid et al. 2002), which could wake up patients. Overall then, we thought that FDI MEPs were a reliable and feasible probe of cortico-spinal excitability in the given experimental setting. For these reasons, although we have no direct evidence for the involvement of the respiratory muscles, our findings represent a new pathophysiological element, indicating a depression of neuromotor descending drive in OSAS, with special reference to apnoea episodes.