CrossTalk opposing view: The intermittent hypoxia attending severe obstructive sleep apnoea does not lead to alterations in brain structure and function

Ivana Rosenzweig (left) is a consultant neuropsychiatrist with research interest in the electrophysiology of sleep and the role of thalamocortical dysrhythmias in neurological and neuropsychiatric illnesses. She shares her research time between the Department of Neuroimaging, Institute of Psychiatry, in London and The Danish Epilepsy Centre in Denmark. Steve Williams (centre) is the Founder, Director and Head of the Centre for Neuroimaging Sciences based at the Institute of Psychiatry and Maudsley Hospital, King's College London. He graduated from Loughborough University in 1985 and spent a year working in high resolution NMR spectroscopy for Beecham Pharmaceuticals in Harlow, UK. As the University of Cambridge's first PhD in MRI, he went on to set up a University of London Intercollegiate Imaging facility at Queen Mary College for the development and application of magnetic resonance techniques in pre-clinical models of disease. In 1994, he moved to the Institute of Psychiatry to champion the application of neuroimaging in CNS disorders. Mary Morrell (right) is Professor of Sleep and Respiratory Physiology in the National Heart & Lung Institute at Imperial College London, UK. Following a Fellowship at the University of Wisconsin-Madison, USA, she returned to the UK to set up the Unit of Sleep and Breathing with clinical colleagues at Royal Brompton Hospital, London. Its aim is to investigate the causes and consequences of sleep-related breathing disorders and translate research into improvements in patient care. Her current research focuses on the cardiovascular and neurological impact of sleep apnoea.

Obstructive sleep apnoea (OSA) is defined by repeated nocturnal episodes of pharyngeal obstruction with resultant intermittent hypoxia (IH), reoxygenation and arousals from sleep. These events have been shown to adversely impact multiple brain regions, notably frontoparietal cortices, temporal cortex, hippocampal/parahippocampal region, cerebellum and some white matter tracts (Morrell et al. 2010; Jackson et al. 2011; Kumar et al. 2012). A number of animal studies by Gozal and colleagues have suggested that it is the chronic exposure to IH that leads to oxidant stress, inflammation and related structural and pharmacometabolic changes of the central nervous system (CNS) (for review see Gozal, 2009). These structural changes are further proposed to underlie the neurocognitive and neuropsychiatric deficits found in patients with OSA (e.g. Canessa et al. 2011). However, during the last decade a wider paradigm shift in the way the brain is viewed has occurred. This has led to reconceptualisation of many complex multisystem illnesses that display neurocognitive sequelae, such as OSA.

Today we recognise the brain as a self-organizing system with complex non-linear dynamics in which cognitive processes arise out of the dynamic interaction and oscillations between multiple regions (Jones, 2009). Accordingly, we argue here an additional, hitherto ignored, ‘third-dimension’ to the core symptoms of severe OSA, proposing that their relative uniqueness suggests malfunctioning of the specific large-scale brain circuitry loop.

Multiple deficits can occur in OSA, including depression, disturbances in attention, dysmetria of thought and affect, verbal memory and executive function deficits (Twigg et al. 2010; Jackson et al. 2011; Sforza & Roche, 2012). These symptoms can be conceptualised as a part of the spectrum of two other well-recognised clinical syndromes, thalamocortical dysrhythmia (Llinás et al. 1999) and the cerebellar cognitive affective syndrome (Schmahmann & Sherman, 1998). Llinás and colleagues (1999) suggest that the endophenotype of a number of neurological and neuropsychiatric syndromes arises when internally generated, low-frequency network oscillations disrupt the normal state-dependent flow of information between thalamus and cortex. Similarly, Schmahmann & Sherman (1998) proposed a novel clinical entity, the cerebellar cognitive affective syndrome that can arise from disruption of the cerebellar modulation of neural circuits. The neurocognitive symptoms (and indeed neuroanatomical regions of interest) frequently described in OSA patients, in our opinion, share many aspects of these two clinical entities.

Rhythmic entrainment of activity in the interconnected cortico-thalamo-cortico (Jones, 2009) and cerebellar (Ros et al. 2009) networks occurs during all states of consciousness and is reflected in the presence of waveforms at different frequencies in the electro- and magnetoencephalogram (EEG/MEG) (Llinás et al. 1999). Appositely, in a study of apnoeic patients, significant EEG slowing in rapid eye movement (REM) sleep was observed over fronto-central and parietal regions, while EEG slowing during wakefulness was observed over all cortical regions (Morisson et al. 1998). During the non-REM (NREM) sleep, the brainstem and basal forebrain modulatory systems impact on the thalamocortical oscillator weakens. This leads to the generation of predominantly slow oscillations, sleep spindles and other slow rhythms, most of which have been extensively investigated in the fields of memory and brain plasticity, and are seen as crucial to re-establishment of synaptic homeostasis (Frank, 2012; Timofeev et al. 2012; Vukadinovic & Rosenzweig, 2012). Predictably, in OSA, the spindles’ frequency and topography are reported to be disordered, which is indicative of impairment of regulatory homeostatic mechanisms, and/or structural changes in the thalamocortical oscillator (Schönwald et al. 2012).

The intricate and finely tuned interplay of thalamic low-threshold calcium channel conductances probably constitutes the skeleton of most NREM sleep oscillatory activities, which then act to group other fast cortical rhythms such as beta and gamma (Crunelli et al. 2006). Therefore, we hypothesise that both the IH-induced changes in neuromodulatory/neurotransmitter systems (as discussed in Gozal, 2009) and the post-arousal-induced (e.g. via hyperventilation or oxidative stress) deranged excitability of thalamic neurons (Jordan et al. 2011; Nair et al. 2011) eventually combine to produce subacute/chronic changes in the oscillator, and the selective cortical/subcortical networks, which are probably themselves directly targeted by hypoxic injury/inflammation (Cervós-Navarro & Diemer, 1991; Gozal, 2009). Further circumstantial evidence points to the increased risk of developing OSA in epileptic patients (Benbadis & Liu, 2007). Also interestingly, patients with dual diagnosis who were non-compliant with continuous positive airway pressure (CPAP) treatment were shown to be at higher risk of seizures, prompting some to question CPAP as a possible anti-epileptic treatment (Vendrame et al. 2011).

Age-related changes in susceptibility to alterations in brain structure and function resulting from IH

In children with OSA, diminished learning capabilities, increased hyperactivity and incidence of attention deficit disorders have been shown (Gozal, 2009). More recently, an increased propensity for accelerated cognitive decline was implied from data in older people with OSA (Ayalon et al. 2010; Sforza & Roche, 2012). Unique periods of susceptibility during the life span, along with genetic, environmental and lifestyle conditions, appear to determine the severity of IH effects on the CNS (Haddad & Yu, 2009; Sforza & Roche, 2012; Kim et al. 2012). Some of these effects at both ends of the age spectrum possibly also reflect the direct and epigenetic impact of IH during the heterochronus neurodevelopmental myelination process, which in humans, unlike that in other primates, spans several decades and probably contributes to later-life cognitive gains (Bartzokis, 2011; Rosenzweig et al. 2012).

Animal models of OSA

The central role for IH leading to alterations in brain structure is strongly contended by animal models of OSA (for review see Gozal, 2009). However, the results of clinical studies in adults with OSA are less convincing (Morrell & Glasser, 2011; Sforza & Roche, 2012). It is of note that in animal studies both IH and sleep fragmentation (episodic arousal from sleep) have been shown to independently affect similar CNS regions (Nair et al. 2011; Sforza & Roche, 2012) possibly, at least in part, via a nicotinamide adenine dinucleotide phosphate oxidase-dependent pathway (Nair et al. 2011). Moreover, studies of the effects of sleep deprivation on cognition in the general population suggest comparable impairments to those seen in OSA (Killgore, 2010; Jackson et al. 2011). Sleep deprivation studies also imply that the highest impact is on cognitive systems that rely on emotional data (Killgore, 2010). The pathophysiology behind this effect might be indicated in reports showing that manipulation of noradrenaline levels during sleep alters olfactory-based and hippocampal-amygdalar-based learning (Frank, 2012). Given our previous argument, it is of note that the main forebrain source of this neuromodulator is the locus coeruleus and that its activity is shown to be timelocked to the upstate of cortical slow oscillations of NREM sleep (Eschenko et al. 2012).

Conclusion

We have argued that the progressive interplay of both sides of the Janus-faced OSA pathophysiology (i.e. IH and sleep fragmentation) aids structural and pharmacometabolic changes in the thalamocortical oscillator that can eventually translate into decreased cortico-thalamo-cerebellar interaction and a variable clinical presentation on the cerebellar cognitive affective syndrome and thalamocortical dysrhythmia spectrum. The ensuing cortical diaschisis probably underlies some of the cardinal neurocognitive morbidities recorded in susceptible OSA patients. It is hoped that by conceptualizing the alterations in brain structure and function attending severe OSA into one, albeit large, malfunctioning brain-loop (rather than IH vs. sleep fragmentation) we may offer a focused theoretical tool for clinically targeted research of treatment pathways, and future therapeutical interventions.

Call for comments

Readers are invited to give their views on this and the accompanying CrossTalk articles in this issue by submitting a brief comment. Comments may be posted up to 6 weeks after publication of the article, at which point the discussion will close and authors will be invited to submit a ‘final word’.

To submit a comment, go to http://jp.physoc.org/letters/submit/jphysiol;591/2/383