BRAIN 2017: 140; 2066–2078 | 2066 SCIENTIFIC COMMENTARIES A restless night makes for a rising tide of amyloid This scientiﬁc commentary refers to ‘Slow wave sleep disruption increases cerebrospinal ﬂuid amyloid-b levels’ by Ju et al., (doi:10.1093/brain/awx148). As many people reading this commentary will know all too well, our sleep often becomes worse as we age. This is true for both the quantity of sleep we can generate, and the quality of that sleep (Mander et al., 2017). For patients with mild cognitive impairment (MCI) and Alzheimer’s disease, the severity of sleep disruption is markedly greater (Mander et al., 2016; Musiek and Holtzman, 2016). Until recently, sleep disruption was thought of as a symptom of neurodegenerative disease. Now, however, an increasing number of studies indicate that sleep disruption may be a causal and instigating factor linked to the pathophysiology of Alzheimer’s disease. Deﬁcient and poor quality sleep, along with several sleep disorders, predict an increased risk of cognitive decline and the conversion to MCI and Alzheimer’s disease. Sleep impairments precede the onset of such clinical outcomes by years if not decades. Conversely, treating sleep disorders such as sleep apnoea can delay the onset of cognitive decline by almost a decade (Mander et al., 2016; Musiek and Holtzman, 2016). In this issue of Brain, Ju and co-workers add to the evidence linking poor sleep to Alzheimer’s disease by revealing an association between the disruption of non-REM slow wave sleep and CSF levels of amyloid (Ju et al., 2017). Mechanistically, the relationship between sleep and Alzheimer’s disease pathology appears to be bidirectional, as evinced by animal models. Mice engineered to overexpress amyloid-b suffer a selective reduction of, and fragmentation in, non-REM (NREM) sleep (Roh et al., 2012). Conversely, restricting sleep in rodents results in signiﬁcantly higher amyloid-b plaque and tau pathological burden, in part because of wake-related synaptic activity that increases the accumulation of metabolic byproducts, including amyloid-b and tau (Kang et al., 2009; Rothman et al., 2013). The brain’s glymphatic system exhibits higher clearance efﬁciency during NREM sleep relative to wakefulness. This includes the preferential NREM sleep evacuation of amyloid-b, representing a sleep-dependent mechanism that helps avert amyloid-b accumulation (Xie et al., 2013). Therefore, sleep offers a proactive state that reduces extracellular amyloid, while wakefulness represents an active state in which these Alzheimer’s diseaseassociated proteins increase (Fig. 1C). Despite causal manipulations in animal models, existing data linking sleep and Alzheimer’s disease pathology in humans have been largely correlational in nature. In this issue of Brain, Ju et al. directly address this knowledge gap. The study used a powerful experimental approach that combines selective NREM slow wave sleep (SWS) disruption and CSF markers of amyloid-b and tau, as well as markers of sleep/wake regulation (hypocretin-1/orexin A) and inﬂammation (YKL-40). In a within-subject, counterbalanced design, Ju and colleagues directly suppressed NREM SWS using auditory tones, then compared next-day Alzheimer’s disease-related CSF measures to those after a sham (control) night of sleep (Fig. 1A). The core experimental question was simple: does the suppression of NREM SWS trigger an increase in CSF markers of Alzheimer’s disease pathology? Across the experimental night of sleep, the auditory tone perturbation successfully reduced NREM slow wave activity (SWA); an electrophysiological measure of NREM SWS quality. Consistent with the hypothesis, there was a rise in CSF levels of amyloid-b the following day (Fig. 1B). However, this increase in amyloid-b only became signiﬁcant within a subgroup of experimental participants demonstrating the greatest reduction in NREM SWA. This is perhaps not overly surprising considering the small sample size. Also, as predicted, these two experimental measures were signiﬁcantly correlated, such that the magnitude of successful NREM SWA reduction predicted the next-day increase in CSF amyloid-b. The acute NREM SWA reduction did not trigger changes in levels of tau, hypocretin-1, total CSF protein or inﬂammatory markers. However, Ju et al. identiﬁed an additional intriguing set of relationships—relationships that ß The Author (2017). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: email@example.com Scientific Commentaries BRAIN 2017: 140; 2066–2078 | 2067 A B C Figure 1 Schematic of experimental design, results and implications. (A) Experimental protocol. Participants underwent two sessions, each starting with 5–14 days of wristwatch-recorded sleep actigraphy, followed by one night of polysomnography (PSG) recording, with CSF collection the following morning. In one session, non-rapid eye movement (NREM) slow wave sleep (SWS) was impaired across the PSG recording night using auditory tones delivered via headphones whenever slow waves were detected in the PSG recordings (orange arrows). In the other session, the night of PSG sleep involved a sham control condition without auditory tones. (B) Left: Schematic of the experimental impact of NREM SWS suppression relative to the sham control condition on CSF concentrations of amyloid-b (Ab, orange) and tau (yellow). Right: Associations between 6-day habitual sleep efficiency and CSF amyloid-b and tau levels. (C) Left: Theoretical schema of the interacting, reciprocal impact (vicious cycle) between NREM SWS, wake-related and Alzheimer’s disease (AD) pathology. Wake-related metabolic activity leads to increased production of amyloid-b and tau, while greater Alzheimer pathology accumulation disrupts NREM SWS, leading to ever increasing amounts of wakefulness. In contrast, during SWS, clearance of Alzheimer’s disease pathology occurs via the glymphatic pathway, and metabolic production of such pathology is low. Right: Longitudinal dynamics of the ensuing predictive model, wherein ever decreasing sleep quality (time) and electrophysiological sleep quality in midlife leads to greater cumulative pathological burden and increased risk of Alzheimer’s disease. 2068 | BRAIN 2017: 140; 2066–2078 Scientific Commentaries Glossary Glymphatic system: The brain’s waste clearance system. The movement of interstitial fluid throughout the brain results in clearance of waste products that build up from neural activity. NREM slow wave sleep (SWS): Stage of non-rapid eye movement sleep that is characterized by a low-frequency (0.8–4 Hz), high-amplitude synchronized electroencephalogram (called delta waves). This stage of sleep is greatly reduced in ageing. Slow wave activity (SWA): Spectral power in the 0.5–4.5 Hz range during non-rapid eye movement sleep or slow wave sleep. Significant reductions in SWA are observed in middle-aged and older adults, relative to younger adults. are perhaps even more ecologically worrisome. Wristwatch actigraphy measurement of poor habitual sleep quality across the experimental week, termed sleep efﬁciency, predicted higher CSF levels of tau, as well as amyloid-b levels (Fig. 1B). The study therefore identiﬁed two temporally different patterns of relevance—an acute relationship, wherein nightly NREM SWA quality predicted next-day amyloid-b levels (though not tau), and a chronic association, such that sleep efﬁciency across days predicted levels of tau and also amyloid-b. That is, at least two different measures of sleep, across two different time periods, predicted two distinct proﬁles of Alzheimer’s disease-related pathological consequence. More generally, the ﬁndings suggest that impairments in NREM SWS, together with a deterioration in the overall integrity and efﬁciency of sleep, represent speciﬁc (though likely interdependent) factors capable of contributing to Alzheimer’s disease progression (Fig. 1C). Considering that such age-related changes in sleep quantity and quality typically begin early in midlife (Mander et al., 2017), they may even be considered prodromal features of, and instigating triggers underlying, the Alzheimer’s disease pathological cascade. Like all preliminary studies, the work by Ju at al. has limitations. As noted, the overall main effect of SWA disruption increasing CSF amyloid-b was not signiﬁcant when all subjects were considered, but only for a subset of participants. In addition, the auditory tone method used to target SWS disruption leaves open alternative interpretations. The control (sham) condition involved no tones, and thus no awakenings. Furthermore, the auditory tone method did not impact SWS in isolation. REM sleep was also reduced by 30%, and lighter stages of NREM sleep were more abundant. The observed experimental effects could therefore be driven by the stress induced by the experimental paradigm, changes in other sleep stages, or the number of arousals, rather than the speciﬁc loss of SWS itself. Indeed, sleep fragmentation, SWS, and REM sleep have all been associated with measures of amyloid-b and tau burden (Mander et al., 2016). That said, the correlation between amyloid-b and the extent of NREM SWA disruption caused by the auditory tones suggest that these alternative possibilities may be less likely. Nevertheless, future studies should employ a yoked auditory control condition to overcome this issue. In such yoked control conditions, auditory tones could be played irrespective of sleep stage, or during only lighter NREM sleep or REM sleep. Limitations aside, new event horizons emerge because of this work—a testament to its seminal nature. For example, what physiological properties of NREM SWS are critical to the observed causal relationship with overnight changes in CSF amyloid-b? Do the number, morphological shape, frequency, or qualitative size of slow waves bathing the sleeping brain predict overnight clearance or production of soluble amyloid? Is it the degree of synchrony of these waves across the brain that is important, or even the quality of slow waves produced in speciﬁc brain regions most sensitive to amyloid-b and/or tau (Mander et al., 2015)? How should we think about these effects in the framework of sleep as a passive state that simply lacks the presence of detrimental wakefulness, relative to a proactive view of sleep, one in which NREM SWS operates to reduce features of Alzheimer’s disease pathology? At present, evidence favours both being true, rather than a mutually exclusive circumstance (Mander et al., 2016). Discovering answers to these questions is not only important for our mechanistic understanding of disease pathways, but will likely reveal novel targets for therapeutic interventions for those at risk of developing Alzheimer’s disease. Longitudinal studies guided by the report of Ju et al. will now help determine the clinical relevance of sustained increases in CSF amyloid-b as a result of sleep disruption and deﬁciency, signiﬁcantly increasing the risk for developing MCI and Alzheimer’s disease. In other words, is sleep a biomarker capable of forecasting the extent and speed of accumulation of amyloid-b plaques and tau-related neuroﬁbrillary tangles, and thus contribute to Alzheimer’s disease risk. These questions will best be addressed by including neuroimaging measures of Alzheimer’s disease pathology in combination with CSF measures, together with measures of sleep actigraphy, sleep polysomnography, and cognition across midlife and older age. Perhaps even more than the biological implications, the study by Ju et al. should trouble us all when set against the known backdrop of plummeting sleep amounts throughout all industrialized nations. Combined with past work, their study emphasizes a causal relationship between disrupted sleep, amyloid-b and tau pathology in middle-aged humans. That is, poor sleep quality, whether through sleep neglect, age or due to medical illness, has the potential to contribute to the Alzheimer’s disease pathological cascade even in healthy, cognitively intact adults (Fig. 1C). If we are looking to extend not only our lifespan, Scientific Commentaries but also our health span, then prioritizing sleep throughout adulthood seems more sage than ever before. Bryce A. Mander,1 Joseph R. Winer1 and Matthew P. Walker1,2 1 Department of Psychology, University of California, Berkeley, CA 94720-1650, USA 2 Helen Wills Neuroscience Institute, University of California, Berkeley, USA Correspondence to: Matthew P. Walker E-mail: firstname.lastname@example.org doi:10.1093/brain/awx174 References Ju YS, Ooms SJ, Sutphen C, Macauley SL, Zangrilli M, Jerome G, et al. Slow wave BRAIN 2017: 140; 2066–2078 sleep disruption increases cerebrospinal ﬂuid amyloid-beta levels. Brain 2017; 140: 2104–11. Kang JE, Lim MM, Bateman RJ, Lee JJ, Smyth LP, Cirrito JR, et al. Amyloid-beta dynamics are regulated by orexin and the sleep-wake cycle. Science 2009; 326: 1005–7. Mander BA, Marks SM, Vogel JW, Rao V, Lu B, Saletin JM, et al. beta-amyloid disrupts human NREM slow waves and related hippocampus-dependent memory consolidation. Nat Neurosci 2015; 18: 1051–7. Mander BA, Winer JR, Jagust WJ, Walker MP. Sleep: a novel mechanistic pathway, biomarker, and treatment target in the pathology of Alzheimer’s disease? Trends Neurosci 2016; 39: 552–66. Mander BA, Winer JR, Walker MP. Sleep and human aging. Neuron 2017; 94: 19–36. | 2069 Musiek ES, Holtzman DM. Mechanisms linking circadian clocks, sleep, and neurodegeneration. Science 2016; 354: 1004–8. Roh JH, Huang Y, Bero AW, Kasten T, Stewart FR, Bateman RJ, et al. Disruption of the sleep-wake cycle and diurnal ﬂuctuation of beta-amyloid in mice with Alzheimer’s disease pathology. Sci Transl Med 2012; 4: 150ra22. Rothman SM, Herdener N, Frankola KA, Mughal MR, Mattson MP. Chronic mild sleep restriction accentuates contextual memory impairments, and accumulations of cortical Ab and pTau in a mouse model of Alzheimer’s disease. Brain Res 2013; 1529: 200–8. Xie L, Kang H, Xu Q, Chen MJ, Liao Y, Thiyagarajan M, et al. Sleep drives metabolite clearance from the adult brain. Science 2013; 342: 373–7. Correlating quantitative susceptibility mapping with cognitive decline in Alzheimer’s disease This scientiﬁc commentary refers to ‘Cerebral quantitative susceptibility mapping predicts amyloid-b-related cognitive decline’ by Ayton et al., (doi:10.1093/brain/awx137). The association of brain iron and Alzheimer’s disease is frustratingly enigmatic. Iron can participate in so many critical and pathological processes that it is difﬁcult not to ascribe an aetiological role in Alzheimer’s disease, yet direct evidence of its participation remains elusive and indirect evidence through therapeutic targeting contradictory. The function and chronology of iron accumulation are not well understood in part because of difﬁculty in measurement. Prior work by Ayton et al. showed that the concentration of ferritin, an iron storage protein, in the CSF was positively correlated with cognitive decline (Ayton et al., 2015). Apropos of treatments targeting pathological iron, it has been proposed that developing suitable clinical treatments for Alzheimer’s disease will likely require better patient and disease stratiﬁcation (Huang and Mucke, 2012). In this issue of Brain, Ayton and co-workers report the use of iron-sensitive quantitative susceptibility mapping magnetic resonance imaging (QSM-MRI) in tandem with amyloid-b-PET as a potential means of addressing the challenge of access with a non-invasive, high spatial resolution technique (Ayton et al., 2017). In this latest work by Ayton et al., 100 subjects were evaluated for cognitive function on an 18-month basis for 6 years. Of the 100 individuals in the study, 64 were cognitively normal, 17 had mild cognitive impairment, and 19 were previously diagnosed with Alzheimer’s disease as deﬁned by the NINCDS-ADRDA criteria. Amyloid-b-PET scans were performed using the 11C Pittsburgh compound B (11C-PiB) followed by MRI acquisition with T1-weighted and QSM modes. An important distinguishing factor in this study is the stratiﬁcation into groups delineating the presence (Ab + ) or absence (Ab ) of PET-determined amyloid-b pathology. As the authors explain, MRI data from the Ab + groups were most predictive of cognitive decline. For instance, hippocampal QSM was positively correlated with Z-score change in Ab individuals, but negatively correlated in Ab + individuals. In Ab patients, QSM of the temporal lobe was weakly correlated with Z-score change, while that of the frontal lobe was negatively correlated. The association of QSM was region speciﬁc but generally showed a positive correlation in Ab + individuals that the authors hope may predict future cognitive function loss. A key point in the report is that MRI-measured iron was not necessarily associated with cognitive decline unless the individual already had mild cognitive impairment: in individuals with mild cognitive impairment, a higher iron load often correlated with greater cognitive decline. The predictive power of QSM-MRI may be applicable to clinical trials where the Published by Oxford University Press on behalf of the Guarantors of Brain 2017. This work is written by US Government employees and is in the public domain in the United States.