The prevalence of obstructive sleep apnea (OSA) in the general population ranges from 9 to 38% (1). OSA is characterized by repetitive upper airway collapse, which leads to intermittent hypoxia, recurrent arousal responses, and sleep fragmentation (2). OSA is associated with a 1.26-fold risk (95% CI: 1.05–1.50) of cognitive impairment and dementia and has been linked to memory dysfunction (3, 4). One review reported that 11–71% of patients with cognitive impairment have OSA (5). In another study, more than 90% of the enrolled patients with dementia had diagnosed OSA, and 39.1% of the cases of diagnosed OSA were severe (6).

Studies have provided possible explanations of the pathological mechanisms underlying the relationship between OSA and dementia. For example, one study indicated that the respiratory events associated with OSA—namely episodes of apnea and hypopnea—affect cerebral circulation as well as cerebrovascular responses and result in hypercapnia and concomitant hypoxia (7). Intermittent hypoxia is associated with elevated reactive oxygen species formation, which can cause oxidative stress; oxidative stress can lead to inflammatory cytokine activation and, in turn, cerebral neuron impairment (8). In addition, hypercapnia and hypoxia can induce arousal responses, which are associated with fragmented sleep and an increased risk of cognitive impairment (9, 10). Therefore, exploring the effects of the various responses induced by OSA stimuli on individuals' risk of developing neurodegenerative diseases is crucial.

OSA severity is classified using the apnea–hypopnea index (AHI), which is the total number of respiratory events during sleep time. Although it can be used to assess sleep-disordered breathing, the AHI may not be entirely suitable for evaluating all phenotypic subtypes of OSA (11). For example, the AHI does not sufficiently reflect the pattern or level of sleep fragmentation or the effect of arousal responses, although these factors are correlated. Frequent arousal, which leads to sleep fragmentation, can result in neurodegeneration and is associated with dementia (12). Another study recruited male participants and investigated the relationships between brain cortical thickness and sleep parameters measured using polysomnography (PSG) among the participants with severe OSA (13). Those results indicated that the arousal index (ArI) and respiratory arousal index (R-ArI) values of the patients were significantly and negatively correlated with the cortical thickness of the prefrontal and parietal cortex areas, which may elucidate some of the underlying mechanisms of cognitive dysfunction. Additionally, researchers have established numerous animal models to investigate the association between arousal indices and biomarkers of neurodegenerative diseases. In a mouse model of Alzheimer's disease (AD), for example, increased amyloid beta peptide (Aβ) deposition was determined to be related chronic sleep fragmentation induced by intermittent nocturnal arousals (14). In other studies, sleep arousal significantly increased both Aβ and tau protein levels in the interstitial fluid of mice and reduced the clearance efficiency of these proteins in animal models (15, 16). However, the associations among sleep-disordered breathing, arousal responses, and the risk of AD development remain unclear and warrant further investigation.

This explorative study, conducted in Taiwan, investigated the associations between sleep parameters measured using PSG and plasma levels of biomarkers of neurodegenerative diseases. We also compared the sleep parameters of the patients in the low- or high-risk groups, who were grouped on the basis of biomarker levels. These aims to determine the relationships between sleep disorders and the risk of AD development. The findings of this study may help elucidate the effects of the accumulation of neurochemical biomarkers on arousal responses.

Although OSA has been determined to be associated with the formation and accumulation of neurochemical biomarkers, the relationship between the clinical symptoms of OSA, such as intermittent hypoxia and arousal responses, and levels of neurochemical biomarker has not been explored. Therefore, we compared the sleep parameters (measured using PSG) of individuals at low and high risks of developing AD and explored the associations between the participants' sleep parameters and neurochemical biomarker levels. The results reveal that the high-risk group had significantly higher mean values for various indices of sleep-disordered breathing and arousal responses than the low-risk group. We determined that increased risk of developing AD was associated with various arousal response and sleep-disordered breathing indices.

The mean AI, HI, AI_NREM, HI_NREM, ODI-3%, and AHI of the low- and high-risk groups differed significantly despite the lack of significant differences between the OSA severity distributions of the groups. Moreover, significant and positive associations were observed among the AHI, ODI-3%, and computed product. These findings are consistent with those of studies investigating the association between oxygen desaturation and elevated levels of neurochemical biomarkers (23, 24) that serve as indicators of hypoxia, a major risk factor for neurochemical biomarker accumulation. Hypoxia causes neuronal apoptosis and tau hyperphosphorylation (25). One study reported a significant association between high AHI values and neurochemical biomarker levels in American patients with severe OSA without dementia (26). Another prospective study analyzed the sleep disorder characteristics of 298 older women (aged ≥ 65 years) without dementia and reported that an increased oxygen desaturation index (≥15 events/h) was associated with the risk of mild cognitive impairment or dementia after adjustment for age, BMI, and ethnicity (27). Hypercapnia, another key risk factor of OSA, can cause deterioration of the functional and anatomic status of cerebral vessels, which may lead to AD (28). Taken together, the available evidence suggests that sleep-disordered breathing is related to an individual's risk of developing AD.

Regarding arousal responses, the patients in the high-risk group had significantly higher ArI, R-ArI, ArI_NREM, and R-ArI_NREM values than did those in the low-risk group. Moreover, R-ArI and R-ArI_NREM values were positively associated with the increased computed product. These results may be attributed to the pathogenic mechanisms of sleep arousal. Specifically, arousal responses refer to abrupt alterations between sleep and fractional wakefulness (29). Recurrent sleep arousal can interrupt the sleep cycle, alter the sleep architecture, and affect the metabolism of neurodegenerative biomarkers (30). They may also be attributed to the tendency of arousal to disrupt the clearance of neurotoxic proteins, resulting in the increased formation of amyloid plaques and the hyperphosphorylation of tau protein strands in the brain, as demonstrated in another study (31). One review similarly concluded that sleep fragmentation and nighttime awakening were associated with AD progression (32). In the present study, the groups' mean arousal indices measured during NREM sleep differed significantly. This may be explained by the underlying mechanisms of slow-wave sleep, which only occurs in the NREM stage and is associated with the modulation of neurochemical biomarkers; that is, arousals during NREM sleep are likely to interfere with the clearance of neurotoxic proteins. Studies have indicated that lessened or unstable NREM sleep increases the level of neurochemical biomarkers, which is consistent with the findings of the present study (33, 34). Collectively, the results of the present study indicate that the high-risk group had higher mean values for the selected arousal indices because arousal responses may disrupt the clearance of neurotoxic proteins, resulting in sleep fragmentation and, in turn, an increased risk of developing AD.

We further explored the relationships between sleep parameters and the risk of AD development by using logistic regression models with and without adjustment for demographic characteristics. Our findings indicate that frequent respiratory events, intermittent hypoxic episodes, and respiratory arousal responses were significantly associated with an increased risk of developing AD. These findings may be attributable to various risk factors, including hypoxia, oxidative stress, and sleep cycle fragmentation. One review elucidated the pathological roles of hypoxia in AD, which include facilitating the accumulation of neurotoxic proteins, enhancing the hyperphosphorylation of tau protein, diminishing the function of the blood–brain barrier, and accelerating neurodegeneration (35). Research has demonstrated the relationship among AD pathology, oxidative stress, and oxidative damage to the brain (36). Another study reported that sleep discontinuity interfered with the clearance of neurotoxic proteins from the central nervous system by the glymphatic system (37). Similarly, another study suggested that sleep disturbance may cause systemic inflammation, thereby increasing Aβ accumulation, which is thought to be a driver of AD pathogenesis (38). Taken together, the findings of this study reveal that high indices of sleep-disordered breathing and hypoxia and a high frequency of respiratory arousal are associated with an increased risk of developing AD.

The present study has some strengths. First, this study analyzed data derived from patients without cognitive symptoms and observed positive associations among sleep-disordered breathing events, neuron biomarker levels, and AD risk. These outcomes are in accordance with the findings of several studies indicating an association of OSA with AD risk (23). Moreover, in contrast to previous studies analyzing sleep parameters in patients with AD (6, 39), this pilot study focused on patients without any cognitive impairment symptom, and the findings of this study may help in understanding the relationship between OSA and AD risk. Another major finding of this pilot study is the positive associations between the respiratory arousal frequency and neuron biomarker levels. Previous studies have investigated the associations between arousal responses and cognitive impairment or different neuron biomarker plasma levels in children (40, 41). The results of this study demonstrated that respiratory arousal events were associated with elevated neuron biomarker levels and thus increased AD risk in adult patients without cognitive impairment. The findings of this pilot study suggest that sleep-disordered breathing and related arousal responses may increase neuron biomarker levels and thus AD risk.

This study has several limitations that should be addressed. First, although individuals with diagnosed AD were excluded from this study, we did not perform neuropsychological evaluations to assess the brain function of the enrolled participants. Moreover, we did not enroll patients who had cognitive symptoms or were diagnosed as having neurological disorders. The mean age of the enrolled patients was 52.25 years; thus, they may have a relatively low risk of cognitive impairment (42). Nevertheless, future studies should investigate the associations among PSG parameters, cognitive questionnaire responses, and biomarker levels to enhance the robustness of our results. Next, during the recruitment process, the presence of genes associated with neurodegenerative diseases was not determined to be predictive of an individual's risk of developing neurodegenerative diseases. However, the potential effects of genetic factors (e.g., ApoE4) on the participants' baseline levels of neurochemical biomarkers and the risk of AD development must be taken into account (43). We did not measure the levels of biomarkers in the participants' cerebrospinal fluid to compare against their levels in the plasma samples, and only one cross-sectional measurement was performed. We were therefore unable to track shifts in biomarker levels over time or evaluate the potential causal relationship between sleep parameters and individuals' risk of AD development. Next, this study did not measure other biomarkers in plasma, such as p-tau 181, Aβ₄₀, or neurofilament light chain protein. These related biomarkers may help identify the risk of AD and thus the associations between OSA and AD. Although we observed positive associations between sleep-disordered breathing indices and the computed product (Aβ₄₂ × T-Tau), we did not measure the levels of inflammatory biomarkers (i.e., glial fibrillary acidic protein). Examination of the level of this biomarker can help in identifying the relationships among hypoxia, inflammation, and neuron impairment. These limitations should be addressed in long-term follow-up studies involving cohorts of patients with OSA.

Another limitation of this study is that it included a small sample of participants enrolled from a single sleep center. Such a small sample size may affect the generalization of our results to different populations. In addition, we recruited individuals only from a single region in Taiwan as the study population. The relationships between arousal indices and neurochemical biomarkers should be further explored in multicenter studies. Previous studies have indicated that the ratio of slow oscillations in the N3 stage was linked to cognitive impairment or AD risk (44). However, this study calculated only the sleep index during the NREM period to eliminate the first-night effect of PSG instead of splitting them into N1, N2, and N3 stages. The first-night effect, resulting from the sleep laboratory environment or PSG devices, may reduce slow-wave sleep and thus cause the incorrect estimation of sleep parameters in some particular sleep stages (e.g., sleep indices in the N3 stage may be overestimated due to the short N3 period) (45). However, respiratory and arousal events occurring in the N3 stage may be crucial risk factors interrupting neuron biomarker clearance. Therefore, researchers should include a large sample and individuals of different ethnicities as well as perform multiple-night PSG to increase the number of participants and investigate sleep parameters in each sleep stage but with the elimination of the first-night effect.

In this study, drawing on the PSG data and plasma levels of selected biomarkers of a sample population from northern Taiwan, we observed that the group that was at high risk of developing AD (patients with computed products > 382.68 pg²/mL²) had higher mean values for several sleep-disordered breathing and arousal indices (during both the REM and NREM stages) than did the group that was at low risk of developing AD. In addition, higher values for sleep-disordered breathing indices, namely AI, HI, AI_NREM, HI_NREM, AHI, and ODI-3%, were associated with an increased risk of AD. Arousal responses, especially respiratory arousal responses, were also associated with an increased risk of AD. Moreover, sleep-disordered breathing indices (AHI and ODI-3%) and respiratory arousal indices (R-ArI and R-ArI_NREM) were positively associated with the computed product. These results indicate that respiratory events, intermittent hypoxia, and arousal responses, including those that occur during the NREM stage, are associated with an increased risk of developing AD. However, the causal relationships among these factors must be further explored through a longitudinal study.

The raw data supporting the conclusions of this article will be made available by the corresponding author without undue reservation.

The studies involving human participants were reviewed and approved by the Joint Institutional Review Board of Taipei Medical University TMU-JIRB: N201912097. The patients/participants provided their written informed consent to participate in this study.

Conceptualized and designed the study: C-YT, S-MW, and W-TL. Data curation and investigation: Y-CK, C-RH, W-HH, Y-SL, and C-MY. Analyzed the data and drafted the paper: C-YT, S-MW, Y-CK, Y-TL, and MS. Critically revised the manuscript and provided essential intellectual contributions: W-TL, AM, C-MY, DW, H-CL, and C-JW. Project administration: K-YL, J-HK, and W-TL. All authors have approved the final version of the manuscript for publication.

This study was supported by the Ministry of Science and Technology of Taiwan (Grant Number: MOST 111-2634-F-002 -021-) and Taipei Medical University Shuang Ho Hospital (Grant Number: 108TMU-SHH-08). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.