What We Don't Know about Sleep-Related Breathing Disorders in the Elderly

IN THIS ISSUE OF SLEEP, SFORZA, ROCHE, AND COLLEAGUES¹ PRESENT CROSS-SECTIONAL DATA EXAMINING THE ASSOCIATION BETWEEN COGNITIVE function and sleep-related breathing disorders (SRBD) in 827 healthy elderly subjects. An extensive assessment including clinical interview, neurological examination, 12-lead ECG, ECG Holter, and magnetic resonance imaging excluded those with myocardial infarction, previously diagnosed sleep disorders, stroke, dementia, and other cardiac and neurological disorders from the cohort at study entry. Two years after enrollment, 827 (58.5% women, mean age 68 years) of the original cohort had one night of nocturnal unattended at home polygraphy, self-assessed cognitive difficulties in everyday life, depression, anxiety, and daytime sleepiness, and an extensive neuropsychological battery of primarily memory and executive function testing.

Noteworthy findings were: (1) 53.8% prevalence of SRBD, defined as apnea + hypopnea index (AHI) > 15; (2) lack of daytime sleepiness in the entire group with no significant difference in daytime sleepiness between those with and without SRBD; (3) little hypoxemia in those with SRBD, with no difference in minutes SaO₂ < 90% between those with and without SRBD; (4) no difference in anxiety, depression, daytime sleepiness, and cognitive difficulties in everyday life between those with and without SRBD; (5) differences in episodic memory (delayed recall) in those with and without SRBD, with the severe apnea group (AHI > 30) demonstrating the lowest scores, but no significant differences in other cognitive function tests; and (6) in multivariate models no significant or strong relationships between cognitive function and severity of SRBD.¹

This study is unique and important, largely because the findings regarding the relationship between SRBD and cognitive function in healthy older adults were negative on almost every front, with the exception of a consistent deficit in episodic memory. The sample size, strength of the study design, methods, and analysis support the internal consistency and generalizability of the findings. The neuropsychological cognitive function battery was comprehensive, and the measures the investigators used are standard for detecting memory and executive function deficits. Inclusion of another measure of attention, such as the psychomotor vigilance task, may have increased sensitivity to decrements in attention. The approach to data analysis was thoughtful, with targeted bivariate and multivariate analyses to detect differences, if they existed.

So here is what we know: in a large population-based sample of healthy 68-year-olds, a comprehensive measure of cognitive function showed that those with SRBD had the same cognitive function (except for a deficit in delayed recall) as those without SRBD.¹ Yet, much remains unknown about SRBD in the elderly. The biggest unknown is the long-term effect of SRBD on cognitive, everyday function, and other common comorbidities such as heart disease in older adults. It is possible that these healthy older adults without cardiac and neurological disease have recent onset SRBD and are thereby without consequent cognitive impairment of significance. We also do not know the earlier severity of SRBD in this sample, further complicating definitive conclusions about short- and long-term effects of SRBD in the elderly.

Moreover, if this sample of healthy older adults has had SRBD most of their adult lives, what has protected them from common consequences of SRBD? The consequences of SRBD are largely mediated by chronic intermittent hypoxia and sleep fragmentation. Intermittent hypoxia, as seen in sleep disordered breathing, is associated with neural injury and impaired cognitive function.² An animal model testing the causal relationship of long-term intermittent hypoxia, neural damage, and impaired cognitive function was developed and tested in a placebo-controlled trial.³ This model revealed that long-term, intermittent, hypoxia exposure during sleep (i.e., arterial saturation reduction from 96-99% to 73-87% equated to an AHI of 40-60 in humans), in the absence of sleep fragmentation, induced learning/memory impairments. The investigators also identified that in aged rats learning impairment was significantly greater than in younger rats with equivalent hypoxia exposures.⁴ The animal model of SRBD, chronic intermittent hypoxia, and cognitive impairment is consistent with clinical study findings in humans,⁵ but it remains uncertain whether this is the mechanism for the episodic memory deficits in the data of Sforza et al.¹

With evidence that supports intermittent hypoxia, neural damage, and cognitive impairment, why are some SRBD sub-groups apparently unaffected in terms of measurable cognitive outcomes? Trait-like resistance to the cognitive effects of sleep deprivation has been documented in healthy younger adults.⁶ It is possible that some older adults with SRBD have brain compensatory mechanisms that are protective, at least in the early stage of SRBD. Recent neuroimaging advances have enhanced our understanding of the mechanisms underlying variable responses in cognition in SRBD.^7–9 While performing cognitive tasks, brain activation can be measured via blood oxygen level dependent (BOLD) functional magnetic resonance imaging (fMRI). Increased BOLD signal is hypothesized to reflect increased neuronal activity and metabolic demand in specific brain areas, suggesting brain activation or activity with simultaneous performance of a cognitive task. The results of these studies suggest SRBD patients demonstrate (1) different brain areas, not typically activated, with increased brain activation during cognitive task performance compared to healthy matched controls,^8,9(2) cognitive task performance is partially or completely preserved in untreated SRBD patients with increased brain activation^7,8; yet, (3) severity of SRBD (i.e., degree of hypoxia and sleep fragmentation) and duration of SRBD may contribute to less brain activation over time (i.e., compensatory fatigue) and thereby impaired cognitive task performance (i.e., memory and verbal learning) even with treatment.^7,9 Differential brain activation response, or cognitive compensation mechanism, in SRBD of varying severity and duration may explain why cognitive deficits are not universally observed in the SRBD population, particularly in those persons with new-onset SRBD and those with relatively few comorbid conditions (i.e., healthy elders) and modest hypoxia. The sample in Sforza and colleagues study¹ was newly diagnosed, largely asymptomatic of common symptoms of SRBD, relatively healthy, and demonstrated moderate disease with little hypoxia.

Recognizing that older persons with SRBD may differentially experience SRBD in terms of common symptoms, physiologic responses, and measurable outcomes, how can treatment decisions be individualized and specific for the elderly with SRBD? Molecular signatures, the patterns of gene or protein expressions in response to a disease, may provide critical insight to individualized responses, and therefore treatment needs. Arnardottir and colleagues¹⁰ suggest that physiologic processes of SRBD, specifically the changes in molecular pathways, are measurable and may be a molecular signature of both the presence and consequences of SRBD. The specific “down stream” effects of physiologic processes in SRBD can be identified with appropriately timed biomarkers. By assessing specific molecular expressions in relevant biomarkers, it is possible that inter-individual differences may be identified that will guide individualized management of SRBD in the elderly.

Nighttime sleep fragmentation and duration of nighttime sleep, and daytime napping patterns were not measured in this study. A potential explanation for the lack of association of SRBD with cognitive function in this population of healthy older adults is that they had longer sleep duration or took daytime naps and that one or both these protected them from the cognitive consequences of SRBD. Short sleep duration and chronic partial sleep deprivation are associated with hypertension, weight gain, insulin resistance, type 2 diabetes, and increased levels of inflammatory biomarkers such as IL-6 that may affect cognition over time.¹¹ Even a few days of chronic partial sleep deprivation affects cognitive performance, primarily attention.¹² Daytime naps have been shown to improve cognitive function and reduce daytime sleepiness in older adults.¹³

Clearly we do not know the immediate or longitudinal consequences of SRBD in older adults with sleep complaints or early cognitive decline. Mild cognitive impairment, characterized by memory impairment but little or no decline in everyday function, is a transitional stage between normal aging and Alzheimer disease. In a cross sectional study of older women (n = 400; mean age 82.8 years), all sleep-disordered breathing indices, including AHI and measures of hypoxia, were associated with cognitive impairment as measured by the Mini-Mental State Exam [AHI of 15.5: odds ratio (OR) 1.4, 95% confidence interval (CI) = 1.03-1.9; AHI of ≥ 30: OR 3.4, 95% CI = 1.4-8.1; and SaO₂ nadir < 80%: OR 2.7, 95% CI = 1.1-6.6].¹⁴ Each standard deviation increase in AHI was associated with 70% greater odds of cognitive impairment. Interestingly, recent studies using neuroimaging in persons with SRBD have shown evidence of hippocampal atrophy and reduced gray matter concentrations in the hippocampus, as well as the frontal and parietal cortices, temporal lobe, and cerebellum,^15–17 although these results have not always been replicated.¹⁸ Other neuroimaging findings in SRBD include reduced cerebral blood flow during wakefulness and a correlation between the severity of SRBD and the prevalence of silent cerebrovascular lesions.¹⁹ In another study, 123 subjects with APOE4 allele and SRBD were compared to 123 controls with APOE4, but without SRBD. There was significantly greater verbal and working memory impairment in the APOE4 plus SRBD patients. The findings from this study provide evidence for significant interactive effects of APOE4 allele and SRBD on cognitive function in older adults and support the possibility of a relationship between SRBD and Alzheimer's disease.²⁰

Future research on SRBD and cognition should examine changes in molecular domains affected by SRBD, such as sympathetic activity, oxidative stress, inflammation, and neuronal changes to potentially predict who is likely to develop OSA-related cognitive decline. This research will importantly guide phenotyping of elders with SRBD, providing critical empiric insight to, at least in part, address what we don't know about SRBD in the elderly and further refine what we do know about SRBD in the elderly.

DISCLOSURE STATEMENT

The authors have indicated no financial conflicts of interest.