Nocturnal Hypoxemia Is Associated with Altered Parahippocampal Functional Brain Connectivity in Older Adults at Risk for Dementia

Abstract

Background:

Obstructive sleep apnea is associated with an increased risk of developing mild cognitive impairment and dementia. Intermittent nocturnal hypoxemia in obstructive sleep apnea is associated with brain changes in key regions that underpin memory.

Objective:

To determine whether older adults with severe nocturnal hypoxemia would exhibit reduced functional connectivity within these regions, with associated deficits in memory.

Methods:

Seventy-two participants 51 years and over underwent polysomnography with continuous blood oxygen saturation recorded via oximetry. The oxygen desaturation index (ODI, 3% dips in oxygen levels per hour) was the primary outcome measure. ODI was split into tertiles, with analyses comparing the lowest and highest tertiles (N = 48). Thirty-five of the 48 participants from these two tertiles had mild cognitive impairment. Participants also underwent resting-state fMRI and comprehensive neuropsychological, medical, and psychiatric assessment.

Results:

The highest ODI tertile group demonstrated significantly reduced connectivity between the left and right parahippocampal cortex, relative to the lowest ODI tertile group (t(42) = –3.26, p = 0.041, beta = –1.99).The highest ODI tertile group also had poorer working memory performance. In the highest ODI tertile group only, higher left-right parahippocampal functional connectivity was associated with poorer visual memory recall (between-groups z = –2.93, p = 0.0034).

Conclusions:

Older adults with severe nocturnal hypoxemia demonstrate impaired functional connectivity in medial temporal structures, key regions involved in sleep memory processing and implicated in dementia pathophysiology. Oxygen desaturation and functional connectivity in these individuals each relate to cognitive performance. Research is now required to further elucidate these findings.

Keywords

Alzheimer’s disease default mode network memory mild cognitive impairment obstructive sleep apnea

INTRODUCTION

It is well documented that obstructive sleep apnea (OSA) is associated with cognitive impairment. In middle-aged OSA adults, neuropsychological deficits across executive functions, attention/processing speed, and memory have been demonstrated [1, 2]. In aging cohorts, the association between OSA and cognition has also been demonstrated within the domain of memory, with the magnitude of the effect more pronounced in clinical, as compared to community, cohorts [3]. Yaffe and colleagues [4] published seminal work demonstrating that OSA (defined by an apnea hypopnea index (AHI) greater than 15) was associated with an increased risk of developing mild cognitive impairment (MCI) and dementia over a median of five years follow-up. Typically, the amnestic subtype of MCI (aMCI, i.e., cognitive deficits in at least the memory domain) is strongly linked with the development of Alzheimer’s disease (AD) [5], while the non-amnestic subtype (naMCI, i.e., deficits in other domains, such as executive functioning or processing speed) may show more diverse disease trajectories [6]. A meta-analysis of prospective cohort studies also demonstrated that those with OSA were 26% more likely to develop cognitive impairment [7]. In a recent study of cognitively normal older adults, OSA was also associated with increased amyloid burden over a two-year follow-up, supporting the hypothesis that OSA is a significant risk factor for developing AD [8].

While the precise mechanisms by which OSA contributes to dementia are unknown, features such as sleep fragmentation and intermittent hypoxia are believed to be key etiological mechanisms. Although AHI severity alone appears to be a poor predictor of cognition in both middle-aged and older samples [2 , 10], nocturnal hypoxemia is likely to be relevant. The oxygen desaturation index (ODI) is a commonly used measure of intermittent hypoxia that quantifies the number of dips in oxygen saturation level by 3% or more per hour of sleep. Higher ODI values have been associated with greater risk of cognitive decline in aging, for both men [11] and women [4].

Neuroimaging studies in older adults have shown that nocturnal hypoxemia is associated with whole brain white matter atrophy, grey matter atrophy in the parietal lobe, and reduced hippocampal volumes independent of covariates including AHI [12]. In an older sample with cognitive concerns, we recently showed [13] that markers of nocturnal hypoxemia were associated with cortical thinning in the brain’s temporal cortex bilaterally, which, in turn, was associated with poorer performance on tests of verbal memory encoding.

In light of the inter-relationship between OSA, cognitive deficits, and structural brain changes, it is plausible that nocturnal hypoxemia may also be pathophysiologically linked to the functional connectivity of the brain, both during wake as well as during sleep. Indeed, in middle-aged OSA samples, studies have variously shown decreased functional connectivity in frontal, temporal, and parietal regions, in addition to increased connectivity in thalamic, cerebellar, and sensorimotor areas [14 –16]. Deactivation of the brain’s default mode network (DMN) has also been reported [17]. The DMN is a network of brain regions that are preferentially yet not exclusively activated during the resting-state [18, 19], and this network appears to be disrupted in regions including the parahippocampal cortex and the hippocampal formation, in those with MCI and sleep disturbances relative to those with MCI without sleep disturbances [20]. Li and colleagues [21] reported that male patients with severe OSA had decreased connectivity in the DMN between the hippocampal formation and the posterior cingulate, medial prefrontal cortex, and the left medial temporal lobe. In OSA, areas of altered functional connectivity have been linked to cognitive dysfunction as well as increasing AHI [22, 23], but to date no known studies have examined the DMN in relation to nocturnal hypoxemia.

While the majority of work exploring how OSA relates to functional dysconnectivity has been conducted in middle-aged or community cohorts, a specific focus on older adults at-risk for dementia but not seeking treatment for symptoms of sleepiness or sleep disturbance is now warranted for a number of reasons. First, individuals with age-related disease such as MCI and AD demonstrate altered functional connectivity within the DMN, particularly in the posterior cingulate/precuneus and hippocampal regions [24, 25]. Second, in older adults, OSA may be driven by different mechanisms (e.g., changes in the upper airway, medical comorbidities, medication, neurodegenerative disease) and it is unclear whether the presence and degree of OSA differentially affects functional brain connectivity. Finally, since prior reports have shown that in older adults at-risk of dementia, oxygen desaturation is associated with structural brain changes in key regions that underpin memory, it is now prudent to investigate whether oxygen desaturation is also associated with functional brain dysconnectivity. Unveiling these inter-relationships will ultimately enable greater scientific understanding of the cognitive neuroscience of OSA, but more importantly will inform strategies aimed at earlier detection of OSA and early intervention for cognitive decline leading to dementia.

In this study, we primarily sought to examine whether in older adults at-risk for dementia, functional connectivity in the DMN differs in those with minimal compared to severe nocturnal oxygen desaturation. We hypothesized that those with substantial oxygen desaturation would exhibit reduced functional connectivity in the DMN, in regions involving medial temporal structures. In addition, we examined how such groups may differ on key measures linked to the DMN temporal regions including memory and working memory. As an exploratory aim, we also examined in these groups how the strength of functional connectivity within the DMN is correlated with neuropsychological performance.

MATERIAL AND METHODS

Participants

Seventy-two middle-aged to older adults (mean age = 66.6, sd = 7.9, range = 51–86; 64% female) at-risk for dementia were recruited from the Healthy Brain Ageing Program at the Brain and Mind Centre, Sydney, Australia. As described previously [13], ‘at-risk’ was defined as those seeking assessment for cognitive decline, including those with subjective and/or objective cognitive complaints. Exclusion criteria were: prior stroke, transient ischemic attack, or head injury (with loss of consciousness >30 min); previous diagnosis of and/or treatment for OSA (e.g., with continuous positive airway pressure); dementia diagnosis or a Mini-Mental State Examination (MMSE [26]) <24; neurological disease (e.g., Parkinson’s disease, epilepsy); psychosis; and inadequate English. This study was approved by the University of Sydney Institutional Ethics Committee, and all participants gave written informed consent prior to study participation. Fifty of the participants in the current study were previously published in a study investigating cortical thickness [13].

Clinical assessment

A medical specialist conducted a physical examination, and recorded medical history, including current medication, using a semi-structured interview. History of psychiatric illness and presence of current major depression was determined using DSM-IV criteria. Additionally, illness burden was assessed using the Cumulative Illness Rating Scale Geriatric Version (CIRS-G) [27] and participants self-reported their current depressive symptoms using the 15-item Geriatric Depression Scale (GDS; 15-item) [28]. The MMSE was conducted for reporting and screening purposes. At the time of assessment, 24 participants (33%) were taking antidepressants, four (6%) were taking adjunctive antipsychotics, two (3%) were taking mood stabilizers, and six (8%) were taking benzodiazepines, with two of these ‘as needed’.

Neuropsychological assessment

A clinical neuropsychologist administered a standardized test battery, as described previously [29]. The overall battery included a comprehensive range of tasks covering an array of cognitive domains. In this study, we were specifically interested in performance on tasks of memory (working memory, verbal and visual memory). Memory was assessed using the Rey Auditory Verbal Learning Test, trial 7 (RAVLT [30]) and the Rey Complex Figure Test (RCFT [30]). Working memory was assessed using the Digit Span subtest of the WAIS-III [31]. Digits forward (DSF) and backward (DSB) were analyzed separately.

MCI diagnoses were consensus-rated by two neuropsychologists and a medical specialist. As per MCI criteria [32], participants were required to demonstrate objective decline (–1.5 SD) on one or more of the cognitive domains assessed by neuropsychological testing. This objective decline was also required to be reported subjectively by the individual and/or their caregiver, with no or minor functional decline. Diagnoses of exclusion also incorporated structural MRI scans where possible. Participants were classified according to the subtype of aMCI if their cognitive deficits included clear impairments in delayed memory recall, or naMCI if their cognitive deficits were in other domains and their delayed memory recall was within normal limits. The sample included 35 participants who met criteria for MCI (n = 9 aMCI and n = 26 naMCI).

Sleep assessment

Participants underwent overnight in-laboratory polysomnography (PSG) in a sleep clinic, which was performed within six weeks of neuropsychological assessment. Nocturnal PSG recordings were collected on an ambulatory recording system (Compumedics Siesta, Melbourne, Vic, Australia; Embla Titanium, Mortara Instruments Inc, Milwaukee, WI, USA). Additionally, nasal airflow using a nasal pressure transducer was recorded, and respiratory effort was assessed using thoracic and abdominal bands. Blood oxygen saturation was recorded via finger pulse oximetry.

Participants were required to maintain their normal bedtime and wake-up schedule for the overnight sleep study, and were asked to refrain from consuming caffeinated beverages for eight hours prior to and during PSG data collection. Sleep architecture and respiratory events were manually scored in 30 s epochs by an experienced sleep technician using standardized scoring criteria [33]. An accredited sleep physician then reported on the studies. The primary outcome measure for this study was the ODI (number of dips in oxygen saturation ≥3% per hour of sleep). Apnea hypopnea index (total number of apneas + hypopneas - reduced airflow ≥30% for ≥10 s, with arousal or ≥3% desaturation - per hour of total sleep; AHI) was also calculated.

To compare those with no or minimal oxygen desaturation to those with significant nocturnal desaturation, the ODI variable was sorted in ascending order, then split into three tertiles. Tertiles divide an ordered distribution into three sections, with each containing a third of the sample. Accordingly, each tertile had n = 24 participants (ODI–high group: mean = 33.95 events/h, sd = 24.1, range =10.80–109.70; ODI–middle group: mean = 7.23 events/h, sd = 1.7, range = 4.83–10.50; ODI–low group: mean = 1.72 events/h, sd = 0.9, range = 0.20–3.70). Subsequent analyses undertaken compared the highest tertile with the lowest tertile only.

Image acquisition and analysis

Acquisition

Participants underwent resting-state MRI, performed within four weeks of PSG assessment. MR images were acquired on a GE MR750 3-Tesla MRI (General Electric, Milwaukee, USA). T2-weighted echo planar images (EPI) were acquired with TR = 3000 ms, TE = 36 ms, flip angle = 90°, 39 axial slices covering the whole brain, field of view = 240 mm, and in-plane voxel size = 3.75 mm×3.75 mm×3 mm, matrix: 64×64, total scan time = 7.25 min. Resting state functional MRI (rsfMRI) data were acquired in a single run, where each participant lay supine in the scanner with their eyes closed and was instructed to allow their mind to wander.

Resting-state fMRI analysis

Image processing and analysis was performed using Statistical Parametric Mapping software (SPM8, Wellcome Trust Centre for Neuroimaging, London, UK) on MATLAB (MATLAB and Statistics Toolbox Release 2015b, The Mathworks Inc., Natick, Massachusetts, US). The first five volume acquisitions of the resting-state EPI data were discarded to eliminate spurious T2-equilibration effects. The remaining 140 scans subsequently underwent pre-processing according to a standardized pipeline. This incorporated slice-time correction to the median (17th) slice within each TR, scan realignment to create a mean realigned image, measures of six-degrees of rigid head movements were calculated for later use in the correction of minor head movements, image normalization (using T2 template) via a 1inear (12-parameter affine) transformation followed by non-linear warping, and Gaussian kernel smoothing (8 mm FWHM isotropic).

Connectivity analysis

After pre-processing, images were imported into the Functional Connectivity toolbox (CONN; http://www.nitrc.org/projects/conn) in MATLAB. A temporal band pass filter was applied (0.009–0.08 Hz) and nuisance parameters (and their first temporal derivatives) were subsequently regressed from the data. These included the six motion parameters extracted from the realignment process and the signal from separate 8 mm ROIs placed within white matter and the cerebrospinal fluid. Furthermore, scans deemed to be movement outliers as detected by the ART toolbox (http://nitrc.org/projects/artifact_detect) were regressed out. Outliers were defined as volumes with frame-wise displacement greater than 0.5 mm or signal intensity changes greater than three standard deviations. These are the recommended conservative settings implemented in the toolbox, based on 95th percentiles in the normative sample. The mean blood oxygenation level dependent (BOLD) signal time courses were then extracted from each ROI. Time courses for each ROI were then correlated with the time courses for all other ROIs, producing Pearson’s r-values that were subsequently converted into z-scores using a Fisher’s r-to-z transformation.

Default mode network statistical analysis

An ROI-to-ROI analysis was conducted, with 8 mm spherical ROIs selected according to those which comprise the DMN, based on previously published coordinates [18]. The 20 ROIs incorporated the dorsomedial and ventromedial prefrontal cortex, as well as bihemispheric regions of: the anterior medial prefrontal cortex, posterior cingulate cortex, temporal pole, lateral temporal cortex, retrosplenial cortex, hippocampal formation, posterior inferior parietal lobule, temporoparietal junction, and the parahippocampal cortex. Unpaired t-tests were carried out on each resting-state DMN connection to determine whether there were any between-groups differences. The significance threshold was determined using a false discovery rate (FDR) with α set to 0.05 [34]. p-values reported throughout this manuscript with respect to functional connectivity analyses are the FDR-corrected values.

Statistical analyses

Demographic data were analyzed using the Statistical Package for Social Sciences (SPSS version 24, Armonk, NY: IBM Corp.). Independent samples t-tests were conducted for group comparisons. Where variable non-normality was present, nonparametric Mann Whitney U tests were implemented. Relationships between categorical variables were assessed via Pearson’s chi-square tests, or Fisher’s exact tests where group sizes were small. Analyses were two-tailed with an alpha level of 0.05. As previously indicated, analyses proceeded comparing the highest and lowest ODI tertiles (n = 24 in each group). In order to account for variance in the connectivity values that may be associated with the potential confound of age, this was included as a covariate in the model. Partial correlation coefficients were used to assess relationships between cognition, functional connectivity, and ODI tertile group (ODI-high versus ODI-low), controlling for age and BMI. Given that MCI was detected in 19 out of 24 in the ODI-low group, as well as 16 out of 24 in the ODI-high group, MCI was also included as a control variable so that specific relationships between oxygen desaturation and DMN connectivity could be assessed with greater specificity by taking into account cognitive heterogeneity within groups.

We conducted partial correlations to assess the relationship between ODI group and cognitive variables, while controlling for the influence of relevant covariates (i.e., age and BMI), in order to reveal associations between ODI group and each of the cognitive variables above and beyond any effects of these covariates. Age was controlled for as the cognitive variables used were raw scores (i.e., had not been adjusted for age via using normative data). This increases the validity and interpretability of the ODI group-cognitive variable relationships. BMI was controlled for in these analyses as higher oxygen desaturation is known to be associated with BMI, and as stated above we wanted to control for this effect so that we could get a better estimation of the relationship between oxygen desaturation and cognitive performance only. BMI was not able to be included as a further covariate in the functional connectivity analyses due to a lack of statistical power. We also did not have hypotheses about BMI affecting connectivity in the DMN in this sample. However, this may be an interesting avenue of inquiry that future, larger studies can address.

RESULTS

Demographics

As shown in Table 1, there were no significant differences between ODI-low and ODI-high groups on measures of age, sex, education, cognition, medical burden, subjective sleep quality, percentage of those with current major depression, age of depression onset, lifetime depression history, MCI diagnosis or subtype, antidepressant, antipsychotic, mood stabilizer, or benzodiazepine usage.

Table 1

Descriptive data (M, SD) of ODI groups

Measure	ODI – low (n = 24)	ODI – middle (n = 24)	ODI – high (n = 24)	p
Gender, % female (n)	67 (16)	67 (16)	59 (14)	0.55
Age	65.88 (8.3)	67.00 (8.2)	66.92 (7.6)	0.65^∧
Education, y	12.57 (3.0)	14.68 (2.1)	12.60 (3.1)	0.97^∧
Global Cognition (MMSE)	28.54 (2.3)	28.79 (1.1)	28.67 (1.3)	0.50^#
Oxygen Desaturation Index	1.72 (0.9)	7.23 (1.7)	33.95 (24.1)	0.00^#
Apnea-Hypopnea Index	8.80 (6.9)	15.26 (15.1)	35.33 (24.9)	0.00^#
Arousal Index	20.45 (23.4)	18.64 (10.8)	33.78 (43.3)	0.00^#
Body mass index	25.89 (4.4)	29.85 (10.4)	29.00 (4.3)	0.01^#
Lifetime depression history, % (n)	50 (12)	46 (11)	58 (14)	0.56
Current depression episode, % (n)	17 (4)	21 (5)	0 (0)	0.11^a
Medical Burden (CIRS-G)	4.70 (5.2)	4.16 (3.1)	4.38 (2.6)	0.71^#
Sleep quality (PSQI)	8.95 (4.0)	7.04 (3.4)	7.38 (3.7)	0.17^∧
Antidepressant use, % (n)	25 (6)	38 (9)	38 (9)	0.40
Antipsychotic use, % (n)	4 (1)	4 (1)	8 (2)	1.00^a
Mood stabilizer use, % (n)	0 (0)	4 (1)	4 (1)	1.00^a
Benzodiazepine use, % (n)	13 (3)	4 (1)	8 (2)	0.41^a
MCI diagnosis, % (n)	79 (19)	83 (20)	67 (16)	0.33
MCI subtype, % amnestic (n)	17 (4)	38 (9)	21 (5)	0.70^a
MCI subtype, % multiple domain (n)	54 (13)	71 (17)	25 (6)	0.12

Tests reported are for group comparisons of ODI-low versus ODI-high only; ODI-middle were not included in the above analyses. All tests are Pearson Chi-square tests unless otherwise specified. ^∧Denotes independent samples t-test. ^#Denotes Mann-Whitney U test. ^aDenotes Fisher’s Exact Test. M, mean; SD, standard deviation; ODI, Oxygen Desaturation Index; MMSE, Mini-Mental State Examination; BMI, body mass index; OSA, obstructive sleep apnea; CIRS-G, Cumulative Illness Rating Scale Geriatric Version; PSQI, Pittsburgh Sleep Quality Index; MCI, mild cognitive impairment.

There were significant group differences between the ODI-low and ODI-high groups on a number of objectively-measured sleep parameters. The ODI-high group had significantly higher AHI (U = 55.00, p < 0.001), number of arousals (U = 188.50, p = 0.040), and ODI (U = 32.50, p < 0.001), relative to the ODI-low group. There were no between-group differences in total sleep time (U = 267.00, p = 0.67) or wake after sleep onset (U = 228.50, p = 0.22). Additionally, as expected, the ODI-high group had significantly higher BMI scores than the ODI-low group (U = 141.50, p = 0.011).

Resting-state connectivity analysis: ODI highest versus lowest tertiles

Controlling for age and MCI diagnosis, analysis of ROI-to-ROI connectivity revealed a significant difference in functional connectivity strength between the highest and lowest ODI tertile groups (see Fig. 1). The ODI-high tertile group demonstrated significantly reduced connectivity between the left and right parahippocampal cortices, relative to the ODI-low tertile group (t(42) = –3.26, p = 0.041 [FDR-corrected], p = 0.002 [uncorrected]; beta (Fisher transformed z) = –1.99, equivalent to r = –0.96). See Table 2 for full network results. As our groups differed on nocturnal awakenings, we repeated this analysis controlling for the amount of wake after sleep onset (WASO). Our results remained the same, with the reduction in functional connectivity between the left and right parahippocampal cortices becoming marginally more statistically significant (t(41) = –3.29, p = 0.039 [FDR-corrected], p = 0.002 [uncorrected]; beta = –2.05). To investigate whether any underlying hippocampal atrophy may be influencing these results, we compared our ODI-high and ODI-low group on left, right, and total hippocampal volumes. All of these were analyses were highly non-significant (t-statistics all <1, p-values all >0.56). To confirm, we repeated the functional connectivity analysis with total hippocampal volume as a covariate. Our results remained the same (t(38) = –3.31, p = 0.039 [FDR-corrected], p = 0.002 [uncorrected]; beta = –2.15).

Fig.1

Significantly diminished functional connectivity ODI-high versus ODI-low; superior view, with spheres denoting PHC left – PHC right. ODI, oxygen desaturation index; PHC, parahippocampal cortex.

Table 2

Default mode network connectivity from the left parahippocampus. Results are controlled for age and MCI diagnosis. Results displayed are connectivity values from the left parahippocampus to all of the other default mode network regions of interest analyzed in this study

Functional connection	Beta (Fisher z)	T-value	p-uncorrected	p-FDR corrected
PHC (left) – PHC (right)	–1.99	–3.26	0.002	0.041
PHC (left) – TPJ (right)	–0.83	–1.85	0.071	0.572
PHC (left) – dMPFC	0.71	1.78	0.081	0.572
PHC (left) – LTC (left)	–0.64	–1.32	0.193	0.667
PHC (left) – RSp (left)	0.78	1.24	0.220	0.667
PHC (left) – pIPL (right)	–0.65	–1.24	0.222	0.667
PHC (left) – pIPL (left)	0.52	1.15	0.257	0.675
PHC (left) – PCC (right)	0.46	0.88	0.386	0.750
PHC (left) – PCC (left)	0.43	0.87	0.390	0.750
PHC (left) – aMPFC (left)	0.38	0.86	0.393	0.750
PHC (left) – TPJ (left)	–0.18	–0.48	0.630	0.834
PHC (left) – aMPFC (right)	0.22	0.47	0.638	0.834
PHC (left) – HF (right)	0.17	0.39	0.696	0.834
PHC (left) – RSp (right)	0.25	0.37	0.710	0.834
PHC (left) – vMPFC	–0.17	–0.36	0.717	0.834
PHC (left) – HF (left)	0.15	0.32	0.752	0.834
PHC (left) – LTC (right)	0.15	0.31	0.755	0.834
PHC (left) – TempP (left)	0.02	0.04	0.968	0.983
PHC (left) – TempP (right)	0.01	0.02	0.983	0.983

PHC, parahippocampal cortex; TPJ, temporoparietal junction; dMPFC, dorsomedial prefrontal cortex; LTC, lateral temporal cortex; RSp, retrosplenial cortex; pIPL, posterior inferior parietal lobule; PCC, posterior cingulate cortex; aMPFC, anteromedial prefrontal cortex; HF, hippocampal formation; vMPFC, ventromedial prefrontal cortex; TempP, temporal pole.

Associations with cognitive function and functional connectivity

Using partial correlations to control for age and BMI (see Table 3), the highest ODI tertile group was significantly associated with poorer digits span forward (DSF) total score, as well as a poorer DSF maximum span. However, groups did not differ on memory learning or recall indicating that ODI group membership was not related to verbal memory performance during wakefulness. Furthermore, ODI group membership was not related to visual memory performance as assessed by RCFT 3-minute recall performance (ODI-low: r = –0.203, p = 0.353; ODI-high: r = –0.177, p = 0.430).

Table 3

Correlations between cognitive, sleep and functional connectivity variables

Groups	Variable	r	df	p
ODI-low versus ODI-high (controlling for age and BMI)	DSF total	–0.345	44	0.019
	DSF maximum span	–0.376	42	0.012
	RAVLT trials 1–5 total	–0.008	41	0.958
	RAVLT trial 7	–0.062	41	0.694
ODI-high PHC L-R fc	RCFT 3 minute recall	–0.546	20	0.009
ODI-low PHC L-R fc	RCFT 3 minute recall	0.325	21	0.144

ODI, oxygen desaturation index; DSF, digit span forward; PHC L-R fc, parahippocampal cortices left-right functional connectivity; RCFT, Rey Complex Figure Test.

Exploratory bivariate correlations revealed that there was a significant moderate to large association in the highest ODI tertile group between left-right parahippocampal functional connectivity and RCFT 3-minute recall performance (see Fig. 2), with those with higher functional connectivity recalling less (r = –0.546, p = 0.009). In the lowest tertile group, this relationship was not significant (r = 0.325, p = 0.144). A between-groups Fisher’s r-to-z comparison revealed that the difference between the correlations for the two groups was significant (z = –2.93, p = 0.003). When age was included as a covariate, however, partial correlations showed an attenuated relationship (ODI highest tertile group: r = –0.418, p = 0.059; ODI lowest tertile group: r = 0.039, p = 0.862; Fisher’s z = 0.0655, p = 0.131).

Fig.2

Plot of ODI-high and ODI-low groups by PHC functional connectivity values with RCFT 3-minute recall performance. ODI, oxygen desaturation index; PHC, parahippocampal cortex; RCFT, Rey complex figure test.

DISCUSSION

This study examines nocturnal hypoxemia and functional brain connectivity in a clinical cohort of older adults with primary cognitive complaints who were not seeking assessment or treatment for any sleep concerns such as OSA. We sought to determine whether those with a high oxygen desaturation index (ODI) demonstrated functional connectivity deficits in the DMN, relative to those with a low ODI. Our data demonstrate that those with a severe ODI (average of 33.95 desaturations ≥3% per hour of sleep, range 10.80–109.70) show reduced functional brain connectivity between the left and right parahippocampal cortices. These findings align with our hypothesis that those with substantial oxygen desaturation would exhibit reduced DMN functional connectivity in regions involving medial temporal circuitry. We also found that greater levels of oxygen desaturation are associated with poorer working memory performance, and that within those with the most pronounced oxygen desaturation, parahippocampal functional connectivity was associated with poorer visual memory recall.

The parahippocampus is a key component of the extra-medial temporal lobe regions and is critically linked to the episodic memory neural networks [35, 36]. Functional neuroimaging data suggests that this region plays a major role in retrieval of factual knowledge [37], episodic future thinking [38], balanced time perspective [39], and maintenance of visual information relative to temporal order [40]. This region has been implicated in the pathophysiology of MCI and AD [41, 42], and recent studies have also shown that the parahippocampus is affected in subcortical vascular MCI [43]. In addition to being linked pathophysiologically to early-stage dementia, this region may also play a key role in sleep-dependent memory consolidation. The parahippocampus is activated and synchronized with hippocampal ripples during attentionally-demanding tasks suggesting that this region plays a critical role in information transfer [44]. The parahippocampal cortex also appears to play an important role in spatial memory [45], with episodic/spatial memory formation presumed to involve bidirectional connections between the neocortex, hippocampus, and parahippocampal region [46]. Reduced parahippocampal activation following sleep deprivation has also been associated with subsequent episodic memory impairment [47]. Previous research in humans has shown the parahippocampal cortex to be preferentially activated, relative to the hippocampal formation, during a visual associative recognition memory task, suggesting that it plays a functionally distinct role in long-term coding of associative relationships [48]. As such, our results demonstrating reduced functional connectivity between the parahippocampal cortex in those with severe oxygen desaturation may indicate disruption of the integrity of temporal lobe structures that are crucial for maintaining effective daytime and nocturnal memory system functioning.

Moreover, our findings revealed that within those with the most pronounced oxygen desaturation, parahippocampal functional connectivity was associated with poorer visual memory recall. However, it is intriguing that our data in those with severe oxygen desaturation revealed that greater connectivity was associated with poorer visual memory performance, a somewhat unexpected finding. It is possible that this relationship reflects a type of compensatory activation. For instance, in MCI, compensatory increases in functional connectivity have been observed, with regions having increased hippocampal connectivity demonstrating negative correlations with episodic memory performance [49]. This may reflect recruitment of additional brain resources (in an attempt to compensate for cognitive functionality losses in other brain regions). Notably, such increases are thought to be temporary and weaken once pathology becomes more widespread [50]. Thus, in our at-risk sample, such increased activation may be similarly compensatory, whereby the parahippocampal cortex may have increased activation due to other brain regions becoming compromised from oxygen desaturation. This is consistent with the lack of such a finding in our group with minimal levels of oxygen desaturation. It should be noted that this finding attenuates when controlling for age, indicating that this effect is at least in part mediated by age. Nevertheless, the effect approaches significance (p = 0.059), and it is possible that this change from statistically significant to borderline has more to do with a lack of statistical power, i.e., small group size (n = 24) coupled with the introduction of an additional covariate (age). Future studies with larger group sizes and/or more established disease are needed to confirm/negate and/or further explore this finding.

In addition to reduced parahippocampal functional connectivity, participants with severe oxygen desaturation performed significantly worse on a task assessing working memory relative to the minimal oxygen desaturation group. If indeed compensatory mechanisms are operative, it may be that brain structures underpinning working memory performance (such as the prefrontal cortex [51]) are linked to compensatory parahippocampal activation. Our prior work examining cortical thinning in those ‘at-risk’ for dementia based on cognitive complaints, has shown that nocturnal oxygen desaturation is associated with thinning in the brain’s temporal cortex bilaterally, which in turn was associated with poorer performance on tests of verbal memory encoding [13]. While highly speculative, it is plausible that brain alterations related to OSA manifest initially as functional dysconnectivity, but with prolonged exposure also demonstrate pronounced structural cortical thinning. In this study, nocturnal hypoxemia was not related to verbal memory performance, and only a small percentage (19%) manifested memory impairment, suggesting that such changes may occur later, or that other mediating factors are relevant. Future studies assessing cortical thinning in association with both episodic memory and functional connectivity changes in the DMN would be useful in delineating these relationships further.

In attempting to delineate how nocturnal desaturation may contribute to alterations in the brain’s structure and function, a number of studies have begun to focus on potential pathophysiological mechanisms. For example, in both human and animal studies, OSA has been shown to be associated with oxidative stress and inflammation and associated neuronal and cognitive decline [52 –54]. These findings suggest that sleep-disordered breathing and nocturnal hypoxemia may meaningfully contribute to brain degeneration, cognitive decline, and subsequent progression to dementia in older adults, and as such may indicate a modifiable risk factor to be targeted for early intervention in at-risk individuals [53]. Specifically, the parahippocampal cortex may be of particular interest when considering cerebral oxygen desaturation. Structural [55, 56] and functional [57, 58] imaging studies have demonstrated alterations in the parahippocampus in association with OSA may be ‘reversed’ with continuous positive airway pressure (CPAP) treatment [59]. Such improvements in parahippocampal alterations associated with OSA with CPAP treatment are postulated to reflect an improvement in muscle sympathetic nerve activity, thought to result from the intermittent periods of hypoxemia that occur during sleep [60]. Indeed, besides its role in cognition, preclinical studies show that the parahippocampus forms part of the higher-order central autonomic network [61] and therefore may play a role in elevated muscle sympathetic outflow [60].

Two recent reports have also shown that cerebrospinal levels of amyloid-β (Aβ₄₂) are altered in OSA [8, 62], suggesting that OSA may directly impact on the development of AD brain pathology. In accordance with our findings, one of these studies also linked Aβ levels to memory impairment in OSA participants [62]. Research has indicated the association between sleep-disordered breathing and cognitive decline is stronger in those carrying the Apolipoprotein E epsilon 4 (APOE ɛ4) gene [63], a variant which is strongly associated with elevated risk for AD [64]. Indeed, cognitive deficits for APOE ɛ4 allele carriers with OSA, relative to non-carriers with OSA, appear to be more pronounced in both spatial working memory [65] and episodic memory [66]. In light of these findings, future studies should look to incorporate Aβ measurement and APOE ɛ4 genotyping when assessing relationships between oxygen desaturation, functional connectivity, and neuropsychological performance. This is of particular importance given the increased risk for developing MCI and dementia incurred by older adults with sleep-disordered breathing [4, 7], and further warranted given that such risk may well be intensified in those with greater amyloid burden and carriers of the APOE ɛ4 allele.

Overall, our findings are innovative in highlighting for the first time that nocturnal oxygen desaturation is linked with changes to the brain’s DMN, specifically to functional dysconnectivity within the parahippocampal brain regions, key regions involved in sleep, cognition, and dementia. Such findings are clinically relevant since they may appear well before the onset of marked verbal memory decline, the most pronounced, albeit small finding arising from meta-analytic studies of cognition and OSA in older samples [3]. Given our sample were health seeking for memory concerns, and were screened for history of OSA, it was also alarming to note the extremely high apnea hypopnea index found in over a third of this sample. Interestingly although our participants may not have met formal AASM criteria for OSA syndrome, our data suggest that hypoxemia due to sleep-disordered breathing should be considered in the assessment of older people with cognitive concerns. This is even more pertinent since there is mounting evidence showing that sleep-disordered breathing is common in older people [67, 68] and is a risk factor for dementia [7]. While traditional screening questionnaires may not work effectively in older people with cognitive impairment [69], increased efforts could now be directed to using simple oximetry screening devices for hypoxemia, which when coupled with effective service pathways, could quickly and effectively target and treat this risk. When coupled with cognitive impairment, nocturnal hypoxemia in particular may have deleterious effects on functional activities such as driving [70]. In addition, nocturnal hypoxemia and sleep fragmentation may even underpin the changes in sleep-dependent memory observed with aging [3] and MCI [71].

In terms of treatment opportunities, while there has been some research examining the effects of CPAP on cognitive function in dementia (e.g., [72, 73]), there are currently a number of trials underway investigating the efficacy and acceptability of CPAP for OSA in MCI (ACTRN12614000442606; NCT03113461; NCT03113461). There have been only a few randomized controlled trials investigating the efficacy of other treatments, such as Mandibular Advancement Splint Therapy, on cognition in OSA (e.g., [74]), suggesting this is a rather neglected area of investigation, particularly for older people. Additionally, it should be noted that other forms of non-pharmacological investigation, such as cognitive training, show promise for improving aspects of cognition [75] and functional brain connectivity [76], and this form of treatment has not yet been examined in those with OSA. Finally, it is important to note that there is considerable heterogeneity in the profile of cognitive deficits in OSA, particularly in determining which individuals may manifest impairment and respond to interventions. This may be at least partially attributed to the duration of untreated disease, as well as other comorbidities and neuroprotective factors such as cognitive reserve [77] and engagement in exercise [78]. To direct personalized medicine, the impacts of such potential mediators should be considered on a case-by-case basis particularly in relation to determining dementia risk and evaluating the most appropriate form of clinical intervention.

One limitation with the present study is the relatively small group sizes. As a result, we did not have the statistical power to comprehensively investigate our findings with respect to oxygen desaturation and cognitive performance, as well as to investigate differences in group-by-PHC connectivity with cognitive performance. A larger study is required to investigate these further and allow for the inclusion of further covariates of interest, as well as potential moderators. Furthermore, the current study is cross-sectional and did not involve direct experimental manipulation or capture longitudinal trends. A longitudinal study in older adults with severe oxygen desaturation would allow further explication of causal and predictive relationships. Assessing both functional connectivity measures and cortical thickness measurements in a single study may help to reveal whether regions demonstrating functional connectivity alterations show corresponding cortical thinning.

Our functional connectivity analyses remained significant after controlling for hippocampal atrophy, suggesting that the functional connectivity deficits were not being driven by underlying atrophy. This is not surprising, as previous studies have indicated that structural brain changes often occur later in the disease process in those undergoing neurodegeneration, and at the MCI or early AD stage atrophy may be insignificant and widely spatially distributed [79]. Furthermore, while temporal atrophy may be relevant to a proportion of those with aMCI, the annual rate of progression from naMCI to AD is approximately 4% [80]. Given that only 9 out of 35 with MCI in our ODI low and high groups had aMCI, and that proportions did not differ between groups, it is unlikely that our results are underpinned by atrophy. It may be that changes in functional connectivity appear earlier and have the potential to be neuroimaging biomarkers of early disease processes.

Overall, our data shows that older adults with high levels of oxygen desaturation demonstrate impaired functional connectivity between left and right hemisphere parahippocampal cortices. Additionally, greater levels of oxygen desaturation are associated with poorer working memory performance. Paradoxically, higher functional connectivity also predicts better visual memory performance in those with the most severe oxygen desaturation, although this may reflect compensatory activation. Further research is now required to further elucidate these findings, with the aim of facilitating effective early intervention and treatment options.

Footnotes

ACKNOWLEDGMENTS

The authors would like to acknowledge the participants who took part in this study. Prof. Naismith is supported by an NHMRC Dementia Leadership Fellowship. Drs. Hoyos, D’Rozario, and Duffy are supported by NHMRC-ARC Dementia Research Development Fellowships. Dr. Shine is supported by an NHMRC Project Grant (GNT1156536) and a Robinson Fellowship from the University of Sydney. Prof. Grunstein holds an NHMRC Senior Principal Research Fellowship (APP1106974). Dr. McKinnon is supported by the NHMRC Centre of Research Excellence to Optimise Sleep in Brain Ageing and Neurodegeneration (CogSleep).

Authors’ disclosures available online ().

References

Stranks

, Crowe

(2016) The cognitive effects of obstructive sleep apnea: An updated meta-analysis. Arch Clin Neuropsychol 31, 186–193.

Wallace

, Bucks

(2013) Memory and obstructive sleep apnea: A meta-analysis. Sleep 36, 203–220.

Cross

, Lampit

, Pye

, Grunstein

, Marshall

, Naismith

(2017) Is obstructive sleep apnoea related to neuropsychological function in healthy older adults? A systematic review and meta-analysis. Neuropsychol Rev 27, 389–402.

Yaffe

, Laffan

, Harrison

, Redline

, Spira

, Ensrud

, Ancoli-Israel

, Stone

(2011) Sleep-disordered breathing, hypoxia, and risk of mild cognitive impairment and dementia in older women. JAMA 306, 613–619.

Levey

, Lah

, Goldstein

, Steenland

, Bliwise

(2006) Mild cognitive impairment: An opportunity to identify patients at high risk for progression to Alzheimer’s disease. Clin Ther 28, 991–1001.

Petersen

, Roberts

, Knopman

, Boeve

, Geda

, Ivnik

, Smith

, Jack

(2009) Mild cognitive impairment: Ten years later. Arch Neurol 66, 1447–1455.

Leng

, McEvoy

, Allen

, Yaffe

(2017) Association of sleep-disordered breathing with cognitive function and risk of cognitive impairment: A systematic review and meta-analysis. JAMA Neurol 74, 1237–1245.

Sharma

, Varga

, Bubu

, Pirraglia

, Kam

, Parekh

, Wohlleber

, Miller

, Andrade

, Lewis

, Tweardy

(2018) Obstructive sleep apnea severity affects amyloid burden in cognitively normal elderly. A longitudinal study. Am J Respir Crit Care Med 197, 933–943.

Terpening

, Lewis

, Yee

, Grunstein

, Hickie

, Naismith

(2015) Association between sleep-disordered breathing and neuropsychological performance in older adults with mild cognitive impairment. J Alzheimers Dis 46, 157–165.

10.

Kushida

, Nichols

, Holmes

, Quan

, Walsh

, Gottlieb

, Simon

Jr , Guilleminault

, White

, Goodwin

, Schweitzer

(2012) Effects of continuous positive airway pressure on neurocognitive function in obstructive sleep apnea patients: The Apnea Positive Pressure Long-term Efficacy Study (APPLES). Sleep 35, 1593–1602.

11.

Blackwell

, Yaffe

, Laffan

, Redline

, Ancoli-Israel

, Ensrud

, Song

, Stone

(2015) Associations between sleep-disordered breathing, nocturnal hypoxemia, and subsequent cognitive decline in older community-dwelling men: The osteoporotic fractures in men sleep study. J Am Geriatr Soc 63, 453–461.

12.

Zuurbier

, Vernooij

, Luik

, Kocevska

, Hofman

, Whitmore

, Ikram

, Tiemeier

(2016) Apnea-hypopnea index, nocturnal arousals, oxygen desaturation and structural brain changes: A population-based study. Neurobiol Sleep Circadian Rhythms 1, 1–7.

13.

Cross

, Memarian

, Duffy

, Paquola

, LaMonica

, D’Rozario

, Lewis

, Hickie

, Grunstein

, Nasmith

(2018) Structural brain correlates of obstructive sleep apnoea in older adults at risk for dementia. Eur Respir J 52, 1800740.

14.

Peng

D-C

, Dai

X-J

, Gong

H-H

, Li

H-J

, Nie

, Zhang

(2014) Altered intrinsic regional brain activity in male patients with severe obstructive sleep apnea: A resting-state functional magnetic resonance imaging study. Neuropsychiatr Dis Treat 10, 1819–1826.

15.

Santarnecchi

, Sicilia

, Richiardi

, Vatti

, Polizzotto

, Marino

, Rocchi

, Van De Ville

, Rossi

(2013) Altered cortical and subcortical local coherence in obstructive sleep apnea: A functional magnetic resonance imaging study. J Sleep Res 22, 337–347.

16.

Zhang

, Wang

, Qin

, Li

, Chen

, Zhang

, Yu

(2013) Altered resting-state brain activity in obstructive sleep apnea. Sleep 36, 651–659.

17.

Prilipko

, Huynh

, Schwartz

, Tantrakul

, Kim

, Peralta

, Kushida

, Paiva

, Guilleminault

(2011) Task positive and default mode networks during a parametric working memory task in obstructive sleep apnea patients and healthy controls. Sleep 34, 293–301.

18.

Andrews-Hanna

, Reidler

, Sepulcre

, Poulin

, Buckner

(2010) Functional-anatomic fractionation of the brain’s default network. Neuron 65, 550–562.

19.

Buckner

, Andrews-Hanna

, Schacter

(2008) The brain’s default network. Ann N Y Acad Sci 1124, 1–38.

20.

McKinnon

, Duffy

, Cross

, Terpening

, Grunstein

, Lagopoulos

, Batchelor

, Hickie

, Lewis

, Shine

, Naismith

(2017) Functional connectivity in the default mode network is reduced in association with nocturnal awakening in mild cognitive impairment. J Alzheimers Dis 56, 1373–1384.

21.

H-J

, Nie

, Gong

H-H

, Zhang

, Nie

, Peng

D-C

(2016) Abnormal resting-state functional connectivity within the default mode network subregions in male patients with obstructive sleep apnea. Neuropsychiatr Dis Treat 12, 203–212.

22.

Park

, Palomares

, Woo

, Kang

, Macey

, Yan-Go

, Harper

, Kumar

(2016) Aberrant insular functional network integrity in patients with obstructive sleep apnea. Sleep 39, 989–1000.

23.

Zhang

, Qin

, He

, Li

, Chen

, Zhang

, Yu

(2015) Functional disconnection of the right anterior insula in obstructive sleep apnea. Sleep Med 16, 1062–1070.

24.

Celone

, Calhoun

, Dickerson

, Atri

, Chua

, Miller

, DePeau

, Rentz

, Selkoe

, Blacker

, Albert

(2006) Alterations in memory networks in mild cognitive impairment and Alzheimer’s disease: An independent component analysis. J Neurosci 26, 10222–10231.

25.

Rombouts

, Barkhof

, Goekoop

, Stam

, Scheltens

(2005) Altered resting state networks in mild cognitive impairment and mild Alzheimer’s disease: An fMRI study. Hum Brain Mapp 26, 231–239.

26.

Folstein

, Folstein

, McHugh

(1975) “Mini-mental state”: A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

27.

Miller

, Paradis

, Houck

, Mazumdar

, Stack

, Rifai

, Mulsant

, Reynolds III

(1992) Rating chronic medical illness burden in geropsychiatric practice and research: Application of the Cumulative Illness Rating Scale. Psychiatry Res 41, 237–248.

28.

Sheikh

, Yesavage

(1986) Geriatric Depression Scale (GDS): Recent evidence and development of a shorter version. Clin Gerontol 5, 165–173.

29.

Naismith

, Rogers

, Hickie

, Mackenzie

, Norrie

, Lewis

(2010) Sleep well, think well: Sleep-wake disturbance in mild cognitive impairment. J Geriatr Psychiatry Neurol 23, 123–130.

30.

Rey

(1941) L’examen psychologique dans les cas d’encéphalopathie traumatique (The psychological examination of cases of traumatic encephalopathy). Arch Psychol 28, 286–340.

31.

Wechsler

(1997) Wechsler Adult Intelligence Scale—III, San Antonio, TX: Psychological Corporation. Language and Speech in FIM Craik (Ed), The handbook of aging and cognition, pp. 359–416.

32.

Winblad

, Palmer

, Kivipelto

, Jelic

, Fratiglioni

, Wahlund

, Nordberg

, Bäckman

, Albert

, Almkvist

, Arai

(2004) Mild cognitive impairment–beyond controversies, towards a consensus: Report of the International Working Group on Mild Cognitive Impairment. J Intern Med 256, 240–246.

33.

Iber

(2007) The AASM manual for the scoring of sleep and associated events: Rules. Terminology and Technical Specification.

34.

Benjamini

, Hochberg

(1995) Controlling the false discovery rate: A practical and powerful approach to multiple testing. J R Stat Soc Series B Stat Methodol 57, 289–300.

35.

Hirjak

, Wolf

, Remmele

, Seidl

, Thomann

, Kubera

, Schröder

, Maier-Hein

, Thomann

(2017) Hippocampal formation alterations differently contribute to autobiographic memory deficits in mild cognitive impairment and Alzheimer’s disease. Hippocampus 27, 702–715.

36.

Zhang

, Simon-Vermot

, Caballero

MÁ

, Gesierich

, Taylor

, Duering

, Dichgans

, Ewers

(2016) Enhanced resting-state functional connectivity between core memory-task activation peaks is associated with memory impairment in MCI. Neurobiol Aging 45, 43–49.

37.

Grummich

, Ganslandt

, Buchfelder

, Roessler

(2017) Characterization of a factual knowledge–associated brain memory area by functional magnetic resonance imaging and implementation in tumor and epilepsy brain surgery. World Neurosurg 107, 291–301.

38.

Gilmore

, Nelson

, Chen

H-Y

, McDermott

(2018) Task-related and resting-state fMRI identify distinct networks that preferentially support remembering the past and imagining the future. Neuropsychologia 110, 180–189.

39.

Guo

, Chen

, Feng

(2017) Neural substrates underlying balanced time perspective: A combined voxel-based morphometry and resting-state functional connectivity study. Behav Brain Res 332, 237–242.

40.

Roberts

, Libby

, Inhoff

, Ranganath

(2018) Brain activity related to working memory for temporal order and object information. Behav Brain Res 354, 55–63.

41.

Goerlich

, Votinov

, Dicks

, Ellendt

, Csukly

, Habel

(2017) Neuroanatomical and neuropsychological markers of amnestic MCI: A three-year longitudinal study in individuals unaware of cognitive decline. Front Aging Neurosci 9, 34.

42.

Rasero

, Alonso-Montes

, Diez

, Olabarrieta-Landa

, Remaki

, Escudero

, Mateos

, Bonifazi

, Fernandez

, Arango-Lasprilla

, Stramaglia

(2017) Group-level progressive alterations in brain connectivity patterns revealed by diffusion-tensor brain networks across severity stages in Alzheimer’s disease. Front Aging Neurosci 9, 215.

43.

, Meng

, Wang

, Yang

, Wu

, Zhao

, Wang

(2017) Cerebral gray matter volume reduction in subcortical vascular mild cognitive impairment patients and subcortical vascular dementia patients, and its relation with cognitive deficits. Brain Behav 7, e00745.

44.

Billeke

, Ossandon

, Stockle

, Perrone-Bertolotti

, Kahane

, Lachaux

, Fuentealba

(2017) Brain state-dependent recruitment of high-frequency oscillations in the human hippocampus. Cortex 94, 87–99.

45.

Bohbot

, Kalina

, Stepankova

, Spackova

, Petrides

, Nadel

(1998) Spatial memory deficits in patients with lesions to the right hippocampus and to the right parahippocampal cortex. Neuropsychologia 36, 217–1238.

46.

Eichenbaum

(2000) A cortical–hippocampal system for declarative memory. Nat Rev Neurosci 1, 41–50.

47.

Gujar

, Yoo

S-S

, Hu

, Walker

(2010) The unrested resting brain: Sleep deprivation alters activity within the default-mode network. J Cogn Neurosci 22, 1637–1648.

48.

Düzel

, Habib

, Rotte

, Guderian

, Tulving

, Heinze

H-J

(2003) Human hippocampal and parahippocampal activity during visual associative recognition memory for spatial and nonspatial stimulus configurations. J Neurosci 23, 9439–9444.

49.

Bai

, Zhang

, Watson

, Yu

, Shi

, Yuan

, Zang

, Zhu

, Qian

(2009) Abnormal functional connectivity of hippocampus during episodic memory retrieval processing network in amnestic mild cognitive impairment. Biol Psychiatry 65, 951–958.

50.

Buckner

(2004) Memory and executive function in aging and AD: Multiple factors that cause decline and reserve factors that compensate. Neuron 44, 195–208.

51.

Cabeza

, Daselaar

, Dolcos

, Prince

, Budde

, Nyberg

(2004) Task-independent and task-specific age effects on brain activity during working memory, visual attention and episodic retrieval. Cereb Cortex 14, 364–375.

52.

Duffy

, Lagopoulos

, Hickie

, Diamond

, Graeber

, Lewis

, Naismith

(2014) Glutathione relates to neuropsychological functioning in mild cognitive impairment. Alzheimer Dement 10, 67–75.

53.

Duffy

, Lagopoulos

, Terpening

, Lewis

, Grunstein

, Mowszowski

, Cross

, Hermens

, Hickie

, Naismith

(2016) Association of anterior cingulate glutathione with sleep apnea in older adults at-risk for dementia. Sleep 39, 899–906.

54.

Musiek

, Holtzman

(2016) Mechanisms linking circadian clocks, sleep, and neurodegeneration. Science 354, 1004–1008.

55.

Canessa

, Castronovo

, Cappa

, Aloia

, Marelli

, Falini

, Alemanno

, Ferini-Strambi

(2011) Obstructive sleep apnea: Brain structural changes and neurocognitive function before and after treatment. Am J Respir Crit Care Med 183, 1419–1426.

56.

Macey

, Henderson

, Macey

, Alger

, Frysinger

, Woo

, Harper

, Yan-Go

, Harper

(2002) Brain morphology associated with obstructive sleep apnea. Am J Respir Crit Care Med 166, 1382–1387.

57.

Macey

, Kumar

, Woo

, Yan-Go

, Harper

(2013) Heart rate responses to autonomic challenges in obstructive sleep apnea. PLoS One 8, e76631.

58.

Innes

, Kelly

, Hlavac

, Melzer

, Jones

(2015) Decreased regional cerebral perfusion in moderate-severe obstructive sleep apnoea during wakefulness. Sleep 38, 699–706.

59.

Fatouleh

, Lundblad

, Macey

, McKenzie

, Henderson

, Macefield

(2015) Reversal of functional changes in the brain associated with obstructive sleep apnoea following 6 months of CPAP. Neuroimage Clin 7, 799–806.

60.

Fatouleh

, Hammam

, Lundblad

, Macey

, McKenzie

, Henderson

, Macefield

(2014) Functional and structural changes in the brain associated with the increase in muscle sympathetic nerve activity in obstructive sleep apnoea. Neuroimage Clin 6, 275–283.

61.

Westerhaus

, Loewy

(2001) Central representation of the sympathetic nervous system in the cerebral cortex. Brain Res 903, 117–127.

62.

Liguori

, Mercuri

, Izzi

, Romigi

, Cordella

, Sancesario

, Placidi

(2017) Obstructive sleep apnea is associated with early but possibly modifiable Alzheimer’s disease biomarkers changes. Sleep 40, zsx011.

63.

Bekris

, Yu

C-E

, Bird

, Tsuang

(2010) Genetics of Alzheimer disease. J Geriatr Psychiatry Neurol 23, 213–227.

64.

Spira

, Blackwell

, Stone

, Redline

, Cauley

, Ancoli-Israel

, Yaffe

(2008) Sleep-disordered breathing and cognition in older women. J Am Geriatr Soc 56, 45–50.

65.

Cosentino

, Bosco

, Drago

, Prestianni

, Lanuzza

, Iero

, Tripodi

, Spada

, Toscano

, Caraci

, Ferri

(2008) The APOE ɛ4 allele increases the risk of impaired spatial working memory in obstructive sleep apnea. Sleep Med 9, 831–839.

66.

O’Hara

, Schröder

, Kraemer

, Kryla

, Cao

, Miller

, Schatzberg

, Yesavage

, Murphy

(2005) Nocturnal sleep apnea/hypopnea is associated with lower memory performance in APOE ɛ4 carriers. Neurology 65, 642–644.

67.

Heinzer

, Vat

, Marques-Vidal

, Marti-Soler

, Andries

, Tobback

, Mooser

, Preisig

, Malhotra

, Waeber

, Vollenweider

(2015) Prevalence of sleep-disordered breathing in the general population: The HypnoLaus study. Lancet Respir Med 3, 310–318.

68.

Peppard

, Young

, Barnet

, Palta

, Hagen

, Hla

(2013) Increased prevalence of sleep-disordered breathing in adults. Am J Epidemiol 177, 1006–1014.

69.

Wilson

, Terpening

, Wong

, Grunstein

, Norrie

, Lewis

, Naismith

(2014) Screening for sleep apnoea in mild cognitive impairment: The utility of the multivariable apnoea prediction index. Sleep Disord 2014, 945287.

70.

Cross

, Terpening

, Duffy

, Lewis

, Grunstein

, Wong

, Naismith

(2017) Association between sleep disordered breathing and nighttime driving performance in mild cognitive impairment. J Int Neuropsychol Soc 23, 502–510.

71.

Westerberg

, Lundgren

, Florczak

, Mesulam

, Weintraub

, Zee

, Paller

(2010) Sleep influences the severity of memory disruption in amnestic mild cognitive impairment: Results from sleep self-assessment and continuous activity monitoring. Alzheimer Dis Assoc Disord 24, 325–333.

72.

Ancoli-Israel

, Palmer

, Cooke

, Corey-Bloom

, Fiorentino

, Natarajan

, Liu

, Ayalon

, He

, Loredo

(2008) Cognitive effects of treating obstructive sleep apnea in Alzheimer’s disease: A randomized controlled study. J Am Geriatr Soc 56, 2076–2081.

73.

Cooke

, Ayalon

, Palmer

, Loredo

, Corey-Bloom

, Natarajan

, Liu

, Ancoli-Israel

(2009) Sustained use of CPAP slows deterioration of cognition, sleep, and mood in patients with Alzheimer’s disease and obstructive sleep apnea: A preliminary study. J Clin Sleep Med 5, 305–309.

74.

Naismith

, Winter

, Hickie

, Cistulli

(2005) Effect of oral appliance therapy on neurobehavioral functioning in obstructive sleep apnea: A randomized controlled trial. J Clin Sleep Med 1, 374–380.

75.

Diamond

, Mowszowski

, Cockayne

, Norrie

, Paradise

, Hermens

, Lewis

, Hickie

, Naismith

(2015) Randomized controlled trial of a healthy brain ageing cognitive training program: Effects on memory, mood, and sleep. J Alzheimers Dis 44, 1181–1191.

76.

Pieramico

, Esposito

, Sensi

, Cilli

, Mantini

, Mattei

, Frazzini

, Ciavardelli

, Gatta

, Ferretti

, Romani

(2012) Combination training in aging individuals modifies functional connectivity and cognition, and is potentially affected by dopamine-related genes. PLoS One 7, e43901.

77.

Alchanatis

, Zias

, Deligiorgis

, Amfilochiou

, Dionellis

, Orphanidou

(2005) Sleep apnea-related cognitive deficits and intelligence: An implication of cognitive reserve theory. J Sleep Res 14, 69–75.

78.

Andrade

, Pedrosa

(2016) The role of physical exercise in obstructive sleep apnea. J Bras Pneumol 42, 457–464.

79.

Zhu

, Li

, Terry

, Puente

, Wang

, Shen

, Miller

, Liu

(2014) Connectome-scale assessments of structural and functional connectivity in MCI. Hum Brain Mapp 35, 2911–2923.

80.

Mitchell

, Shiri-Feshki

(2009) Rate of progression of mild cognitive impairment to dementia–meta-analysis of 41 robust inception cohort studies. Acta Psychiatr Scand 119, 252–265.