The Brain in Obstructive Sleep Apnea: the Chickens Coming Home to Roost?

Brain white matter hyperintensities (white matter disease [WMD]) seen on Proton Density, T2 weighted or FLuid Attenuated Inversion Recovery (FLAIR) MR scanning are extremely common with advancing age.^1–3 Typically they occur in periventricular areas or deep white matter (most commonly frontal, less commonly parieto-occipital, and rarely brainstem or basal ganglia). WMD is graded by the extent and confluence of these lesions. Early confluent or confluent changes tend to progress over time^3–5 and are strongly associated with development of stroke, cognitive impairment—especially executive function,^3,5–9—and dementia.¹⁰ The association with stroke persists even after adjustment for conventional vascular risk factors, suggesting either that WMD reflects the total effect of these risk factors better than the presence or absence of each factor alone, or alternatively that other risk factor(s) which are not conventionally recognized may play a role.^3,10 It is tempting to speculate that obstructive sleep apnea (OSA) may be such a missing risk factor. Indeed WMD has been proposed as a useful intermediate marker in the search for risk factors for stroke and dementia.¹⁰ Pathologically these lesions correspond to areas of myelin pallor, gliosis, and tissue rarefaction with myelin and axonal loss, termed “leukoaraiosis”; they are thought to be due to small vessel ischemia.^2,3 Focal lesions tend to have a more diverse etiology³ and have a lower propensity to progress radiologically or to clinical neurological impairment.^5,11

OSA is known to cause many of the risk factors for both stroke and WMD.^2,10,12–14 OSA is also a known risk factor for stroke in both cross-sectional and longitudinal studies,^15–20 and is also associated with cognitive impairment,^21–24 though the literature is not uniformly positive on the latter issue, particularly in the elderly.^25–27 It has been postulated that OSA may also be a risk factor for WMD, and that the presence of WMD may indicate a population of OSA patients at particular risk of stroke and cognitive impairment, who therefore would be a potential target for intensified therapy.

Several neuroimaging studies have previously addressed these questions, with varying results. An early case control study failed to find any association between OSA and WMD.²⁸ Harbison et al. found in 78 acute stroke patients an independent association between WMD severity (presumed to predate the stroke) and apnea-hypopnea index.²⁹ Two publications from the Sleep Heart Health Study failed to demonstrate that OSA was a risk factor for incident WMD in either brainstem³⁰ or cerebral cortex.³¹ In those studies, patients had two MR scans mean (SD) 5.0 (0.6) years apart, with a polysomnogram 3.6 (0.8) years after the first scan. Thus no truly cross-sectional (i.e., simultaneously acquired) data were available, and the timing of any incident WMD in relation to the PSG was uncertain. Only a small number of patients showed deterioration over the period of study, which may have impacted on likelihood of finding associations. However associations were found between arousal index and brainstem WMD progression³⁰ and between central apnea index and cerebral WMD progression.³¹ Subsequently Eguchi et al.³² demonstrated that nocturnal hypoxemia was an independent risk factor for silent cerebrovascular disease in a community sample (n = 146) selected for high cardiovascular risk factors. Cerebrovascular disease included both lacunae and WMD as a composite outcome. Nishibayashi and coworkers³³ found increases in presence and severity of both lacunae and WMD in moderate and severe OSA in 192 middle-aged clinic-referred patients. Most recently Kiernan et al.³⁴ found a high prevalence of WMD in an OSA population (n = 62) but no relationship was found with severity.

The study by Kim and colleagues³⁵ in this month's issue of Sleep is, therefore, the first population-based cross-sectional study to examine the association between OSA and WMD. In 503 subjects aged 40-69 in 2001 at recruitment to the Korean Genome and Epidemiology Study (KoGES), they found a significant increase in WMD across control, mild, and moderate to severe OSA groups. Furthermore they showed a significant increase in severity of OSA across groups with increasing severity of WMD. Moderate to severe OSA was associated with over twice the risk of any WMD compared to a control group with no OSA, after adjustment for relevant covariates. The study is, therefore, a significant step forward in defining an association between the two pathologies.

Nevertheless, the evidence for OSA as a causal factor in WMD in the community remains inconclusive. Despite the strengths of the work of Kim and colleagues, it is not clear how exactly the cohort who had both MR scanning and PSG were derived. It is therefore difficult to know how representative they are of the parent population of almost 11,000 who were randomly telephoned in 2001. In one respect they differed markedly, being 71% female. This may also explain the very low rates of smoking. (In 2009, 44% of Korean males and 7% of Korean females smoked daily.³⁶)

A second issue is the question of the clinical significance of the white matter lesions they observed. Not all white matter hyperintensities carry the same prognosis.³ It is not possible to tell whether a score of 3 represents isolated focal lesions in 3 of the 10 brain regions analyzed, or a single confluent (grade 3)³⁷ lesion in one area.

Lastly this study does not answer the old “chicken or egg” question, to continue the analogy of Gozal³⁸ and of Morrell.³⁹ It remains possible that WMD is the cause rather than the consequence of OSA (i.e., the egg rather than the chicken). The independent association of moderate to severe OSA with presence of WMD after controlling for covariates tends to favor the chicken,³⁵ but this remains an association, and no comparable analysis in search of evidence to favor the egg was reported.

Replication of the findings of Kim et al.³⁵ is now needed, and they are in an ideal position to further investigate the links between OSA and WMD longitudinally in future. Certainly there is a need for further longitudinal data, to determine the influence of OSA on incident WMD and subsequent clinical sequelae, and also the effects of treatment. An Australian population-based study, Snore-ASA (Australian Clinical Trials Registry Number ACTRN12612000891820) is also currently investigating this issue, as well as the influence of aspirin on cognitive decline and WMD progression in OSA and non-OSA populations. Given the high prevalence of stroke, cognitive decline, and OSA in the swelling ranks of the elderly, delineating the potential role of OSA as a modifiable risk factor for WMD and cognitive impairment is of paramount importance. Kim and colleagues have taken an important step in that direction, but much remains to be done.

CITATION

Hamilton GS; O'Donoghue FJ. The brain in obstructive sleep apnea: the chickens coming home to roost? SLEEP 2013;36(5):637-639.

DISCLOSURE STATEMENT

Dr. Hamilton has participated in research studies which have received funding support from Compumedics Ltd and Resmed. Dr. O'Donoghue has participated in research sponsored by Apnex Pty and has participated in speaking engagements for AstraZeneca. He is also a partner in two privately run sleep laboratories.