Progression of Dementia Assessed by Temporal Correlations of Physical Activity: Results From a 3.5-Year, Longitudinal Randomized Controlled Trial

Abstract

Cross-sectional studies show that activity fluctuations in healthy young adults possess robust temporal correlations that become altered with aging, and in dementia and depression. This study was designed to test whether or not within-subject changes of activity correlations (i) track the clinical progression of dementia, (ii) reflect the alterations of depression symptoms in patients with dementia, and (iii) can be manipulated by clinical interventions aimed at stabilizing circadian rhythmicity and improving sleep in dementia, namely timed bright light therapy and melatonin supplementation. We examined 144 patients with dementia (70-96 years old) who were assigned to daily treatment with bright light, bedtime melatonin, both or placebos only in a 3.5-year double-blinded randomized clinical trial. We found that activity correlations at temporal scales <~2 hours significantly decreased over time and that light treatment attenuated the decrease by ~73%. Moreover, the decrease of temporal activity correlations positively correlated with the degrees of cognitive decline and worsening of mood though the associations were relatively weak. These results suggest a mechanistic link between multiscale activity regulation and circadian/sleep function in dementia patients. Whether temporal activity patterns allow unobtrusive, long-term monitoring of dementia progression and mood changes is worth further investigation.

Progression of Dementia Assessed by Temporal Correlations of Physical Activity: Results From a Controlled Trial

OPEN

A

Kun Hu Pa

found that activity correlations at temporal scales <

Dementia and Alzheimers disease (AD) have become increasingly prevalent as the population aged 65 and older continues to increase1,2. To better monitor the onset and progression of dementia, regular assessments of cognitive function and behavior are encouraged, and tools for ambulatory monitoring while not interrupting daily life are desirable3. Here we examined whether a nonlinear measure of activity patterns based on temporal correlations in spontaneous motor activity uctuations can be used to objectively assess cognitive and behavioral changes in patients with dementia.

Human motor activity displays seemingly irregular uctuations over a wide range of time scales from seconds to hours4,5. In healthy young individuals these uctuations are not random but possess robust temporal correlations that are similar at dierent time scales and independent of mean activity levels6,7. Mounting evidence suggests that altered temporal activity uctuations reect the changes in brain function that occur with aging and in diseases5,714. For instance, activity uctuations in older individuals with dementia are more random with reduced correlations at time scales >~2 hours; and the perturbation is more pronounced in AD patients with

Division of Sleep and Circadian Disorders, Brigham and Womens Hospital; and Division of Sleep Medicine, Harvard Netherlands Institute for Neuroscience, Amsterdam, The

SCIENTIFIC REPORTS

1

www.nature.com/scientificreports/

higher cortical plaque density9,10. Recent studies also suggest clinical relevance of activity uctuations at smaller time scales, showing altered activity patterns at <~2hours in patients with mood disorders5,1114. However, these ndings are exclusively based on cross-sectional studies. The clinical relevance of temporal activity regulation in dementia awaits evidence from longitudinal studies that examine within-subject changes during the progression of dementia and the potential association with mood changes.

One possible common pathological pathway for the observed altered multiscale activity regulation in dementia and depression is disrupted function of the sleep/circadian control system15. There are several lines of studies supporting this hypothesis. (1) Cumulative evidence suggests a mechanistic role of sleep/circadian control in the development and progression of dementia16. For instance, disturbances in sleep and circadian rhythms are common in older persons and more pronounced in patients with dementia1719. In addition, recent studies have shown that disturbed sleep and daily activity patterns may be an earliest sign of dementia or AD development2022. (2) Circadian/sleep dysfunction is also associated with mood disturbances. Patients with depression oen have abnormal sleep-wake cycles and reduced circadian amplitudes in daily activity and other physiological functions1719,2325; and treatment of depression with antidepressants helps to restore disrupted sleep and daily/ circadian rhythms18,19. (3) Timed bright light therapy that can improve sleep and SCN functioning is also benecial for mood and cognition26,27. (4) Most related to multiscale activity regulation, functional changes in the circadian pacemaker (suprachiasmatic nucleus; SCN) are accompanied by not only perturbed activity rhythms at ~24hours28,29 but also degraded temporal activity correlations at multiple time scales9,10,30. Despite the evidence, it is unknown whether degraded activity regulation in dementia patients can be attenuated/reversed by manipulation of sleep/circadian regulation.

To better understand the clinical relevance of multiscale activity regulation and the underlying mechanisms, we examined data collected longitudinally to test four hypotheses: (1) temporal activity correlations decrease gradually over time in dementia patients; (2) long-term bright light with melatonin therapy attenuates the degradation in temporal activity correlations over time; (3) the degree of the decline in activity correlations positively correlates with the worsening of depression symptoms in patients/individuals with dementia; and (4) the degree of the decline in activity correlations positively correlates with cognitive declines. To test these hypotheses, we examined motor activity regulation of 165 patients (7096 years old) with mid- to late-stage dementia from an existing database for a double-blind randomized clinical trial (controlled-trials.com/isrctn Identier: ISRCTN93133646; Registered on December 09, 2005)31. These patients were assessed at baseline and every 6 months thereaer for up to 3.5 years. In this clinical trial, 45 patients received daily treatment with bright light, 39 took bedtime melatonin each day, 44 received both the bright-light and melatonin treatments, and 37 received placebos only. To estimate temporal correlations in the activity uctuations at time scales from ~0.1 up to 12hours, we performed the detrended uctuation analysis (DFA) that has been widely used in physiological data analysis15. We determined the within-subject changes in multiscale activity correlations over time and related the results to the changes in mood and cognition of the same subjects.

Results

At baseline (i.e., prior to any intervention) activity uctuations in these dementia patients had positive correlations at all tested time scales (as indicated by DFA exponent >0.5) but the temporal correlations were much weaker (i.e., more random uctuations) at time scales >2hours as compared to that at smaller time scales (DFA exponents: 2=0.720.01 [SE] at>2hours; 1=0.970.01 at<1.5hours;1 2=0.25 0.02; Paired t-test, p< 0.0001) (Figs1 and 2). At baseline, 1 and 2 showed no signicant dierences between the four groups (both p>0.7).

Temporal correlations of activity uctuations at time scales <1.5 hours gradually decreased over the ~3.5 years of follow-up (Fig.1 and 2) with a mean annual reduction of 0.021 0.004 (SE) in 1 (Mixed Model; Time: t ratio = 4.61, p< 0.0001; Fig.2A). In contrast, activity correlations at time scales >2hours (2) showed no signicant change over the period of the study (p>0.8; Fig.2B).

The mixed model revealed an interaction between the eects of time and bright light treatment on 1 (Time Light: 0.024 0.008, 95% CI: 0.00830.039, t ratio=3.02, p=0.0026), i.e., 1 decreased by 0.0330.007 per year in the people who received the placebo light treatment and only by about a third, 0.009 0.004 per year in the people who received the active bright light treatment (Fig.3). These results indicate that bright light treatment counteracted the degradation of short-term activity correlations in dementia by 73% (95% CI: 25118%).

Melatonin treatment did not aect the rate of change in 1 during follow-up (Melatonin Time: p > 0.9) and did not aect the interaction between the eects of time and light treatment (Melatonin Light Time: p > 0.15). There were no signicant eects of light, melatonin, and their interaction on the change of 2 (p values> 0.1 for Light Time, Melatonin Time, MelatoninLightTime).

Similar to 1, cognition, depression, and psychological functions also showed signicant changes from the baseline to the follow-up assessments, indicating functional deterioration over time (Fig.4). At baseline, 1 was significantly associated with the MiniMental State Examination (MMSE) score (18.8 (7.7)1, r=0.25, t ratio=2.5, p = 0.016), the Cornell Scale for Depression in Dementia (CSDD) score (16.5 (7.1) 1, r = 0.24, t ratio =

2.31, p = 0.023), and the multidimensional observation scale for elderly subjects (MOSES) social withdrawal behavior score (19.1 (7.8)1, r= 0.25, t ratio= 2.44, p= 0.017) (Fig.5). These associations indicate that subjects with attenuated activity correlations at small time scales had worse cognition, mood, and social withdrawal behavior.

SCIENTIFIC REPORTS

2

www.nature.com/scientificreports/

Figure 1. Changes in temporal activity correlations over time. (A) Activity recordings of a patient with Alzheimers disease in the group without light or melatonin treatments during baseline, 12-month, and 18-month assessments (B) DFA results of signals in (A). On the abscissa, n represents the time scale in hours. The uctuation functions F(n) are vertically shied for better visualization of changes in the slopes (DFA exponents) over time. The uctuation function was also obtained from surrogate data that were generated from the baseline recording in A by randomly shuffling the data points. Shuffling destroyed the temporal organization of activity uctuations, leading to white noise without correlations as indicated by a power-law function (i.e., a straight line in the log-log plot) with the slope of 0.5 (dashed line) over the entire range of tested time scales.

During the follow-up assessments, mixed models showed that the change in 1 (1: the change from the baseline or the rst assessment) was: (i) positively correlated with the change of MMSE (MMSE~5.6 (2.2)1,

r = 0.11, t ratio = 2.49, p = 0.013); (ii) negatively correlated with the change of CSDD (CSDD~15.2 (3.5) 1, r = 0.19, t ratio = 4.31, p < 0.0001); (iii) marginally associated with the increase in the Philadelphia Geriatric Centre Aect Rating Scale (PGCARS) negative aect scale (PGCARS~3.5 (1.5) 1, r=0.09, t ratio=2.35, p= 0.019); and (iv) negatively correlated with the changes of withdrawn behavior scale of MOSES (MOSES~7.2 (2.5) 1, r = 0.10, t ratio = 2.88, p = 0.0042) (Fig.6). These associations indicate that less reduction in short time scale activity correlations over the years was associated with less cognitive decline, less mood decline and less increase in social withdrawal, respectively. We did not nd any signicant associations between any changes in activity correlation >2hours (2) and the changes of clinical outcome measures (all p values>0.1).

Discussion

Human activity uctuations possess a complex, multiscale temporal organization which is signicantly altered in older individuals and under pathological conditions (e.g., dementia and mood disorders)5,713. The current study provides the rst analysis of within-subject changes of the temporal organization over time, the relevance of such changes to functional declines, and the eect of clinical interventions in people with dementia. Our results show that the temporal activity correlations at <~2hours gradually decline over the years in patients with dementia and that this decline is associated with declines in mood and cognitive function. More importantly, we discovered that bright light treatment attenuated the progressive reduction in activity correlations by ~73%, suggesting for the rst time that it is possible to manipulate human multiscale activity regulation with an intervention. The benecial eect of the light treatment on multiscale activity regulation is consistent with its eects on cognition and depression symptoms. Using the same database, we previously reported that the light treatment attenuated cognitive deterioration by 5% on the MMSE and depression symptoms by 19% on the CSDD31. Since it is believed that the positive health impacts of bright light treatment are via its inuences on SCN functioning and sleep-wake cycles, our ndings support the notion that circadian regulation plays an important role in the

SCIENTIFIC REPORTS

3

www.nature.com/scientificreports/

Figure 2. Correlations in activity uctuations at baseline and during the follow-up assessments.

(A) Correlations at time scales <2hours, as characterized by the DFA exponent 1, decreased gradually over time. (B) Correlations at time scales >2hours, as characterized by the DFA exponent 2, remained virtually unchanged over the period of the study. For better visualization, data during the follow-up assessments were divided into three bins (<6 months, 612 months, and >12 months). P values for the inuences of time onthe exponents were obtained using mixed models in which time into the study was included as a continuous variable (i.e., not binned categorized variable). Mean values were obtained from all individuals in all groups and error bars indicate standard errors (between individuals).

Figure 3. Eects of light treatment on activity correlations. P value is for the eect of the interaction between light and time course on DFA exponents. Data during the follow-up assessments were divided into three bins for better visualization (<6 months, 612 months, and >12 months). Error bars indicate standard errors between individuals.

temporal organization of activity uctuations at multiple time scales15. Together these ndings provide evidence that temporal activity regulation may be clinically relevant.

Frequent assessments of cognitive function and behavior are required to monitor improvement, stabilization or worsening of the disease-related symptoms in dementia patients and to better follow the progression of the disease. Regular assessments are also encouraged even in older persons without previously diagnosed dementia so that appropriate actions can be taken to prevent or delay the onset of dementia32. Our results suggest that the dynamic measure of activity uctuations may serve as a promising tool for a routine assessment of cognition and behavior in older individuals. As compared to traditional clinical methods, this approach has a number of advantages in terms of cost efficiency and exibility. For instance, the measure can enable a continuous assessment when subjects undergo their normal daily activities without the necessity of scheduling a visit with a doctor, physician, or other health care professional. With technical advance in data transmission and storage, collected activity data and results can be reviewed instantaneously or retrospectively to identify dynamic changes in patients conditions from day to day and even at dierent times of day. In addition, the tested activity measure (i.e., temporal correlations) is resilient to external inuences of scheduled behavior and environmental conditions (e.g., daily schedule of interactions between dementia patient and caregivers in a nursing home)6. For instance, we have found that alterations in temporal activity correlations can better reect intrinsic SCN neurodegeneration in dementia when compared to traditional circadian measures such as the amplitude of core body temperature (CBT) rhythm, the amplitude of motor activity rhythm, and intraday variability in motor activity a measure characterizing fragmentation of the activity rhythm10. Thus, the application of this nonlinear measure to the clinic

SCIENTIFIC REPORTS

4

www.nature.com/scientificreports/

Figure 4. Functional outcome measures at baseline and during the follow-up assessments. (A) Mini-mental state examination (MMSE). (B) A subscale of the multidimensional observation scale for elderly subjects (MOSES) for withdrawn behavior. (C) Cornell Scale for Depression in Dementia (CSDD). (D) Negativeaect scale of the Philadelphia Geriatric Centre Aect Rating Scale (PGCARS). Data during the follow-up assessments were divided into three bins for better visualization (<6 months, 612 months, and >12 months). P values were obtained using mixed models in which time into the study was included as a continuous variable and subject as a random factor for intercept. Mean values were obtained from all individuals and error bars indicate standard errors (between individuals).

may potentially simplify monitoring of dementia progression and provide valuable data for a better physiological understanding of the disease and its impact on daily life.

How activity correlations at small time scales (<~12 hours) are altered in mood disorders is still unclear. Three recent studies provided seemingly discrepant results4,12,13. The study of Krane-Gartiser et al. reported more random activity uctuations in acutely admitted inpatients with bipolar disorder12. Though based on dierent analytical tools to examine activity uctuation (i.e., autocorrelation with lag= 1), the nding is consistent with our nding that the reduction of activity correlations at <2hours was associated with more depression symptoms and more impaired cognition in patients/individuals with dementia. However, the two other studies suggested that mood disorders lead to increased temporal correlations in activity uctuations at small time scales. Aybey et al. showed an increased DFA exponent in patients with major depression4 and Sano et al. reported persistence of resting and active periods in Schizophrenia13. The diversity of the deviations from normal is reminiscent of how correlations in heartbeat uctuations are altered dierently under dierent pathological conditions, i.e., heartbeat correlations at >~30 beats are reduced in arterial brillation but are increased in congestive heart failure15,33. It

is possible that varied disorders or dierent stages of a disease may have diering inuences on temporal activity correlations. Thus, further studies are required to clarify the changes of the activity correlations at small time scales for specic phases of specic diseases.

A similar concern applies to activity correlations at larger time scales. We previously found that activity correlations at time scales >~2hours are reduced with aging and in dementia and that the reduction is related to the perturbed circadian regulation9,10,30,34. Since light and melatonin treatments are believed to be benecial for circadian regulation in dementia patients, it would be expected that these treatments may also improve activity correlations at large time scales, but we did not observe such an eect in the current study. We note that most of our participants in this study were in the middle to late stages of AD/dementia (e.g., 88.5% of participants had MMSE 20 and only 19 participants had MMSE >20) and that the baseline activity correlations of the participants at large time scales were already signicantly reduced (Figs1 and 2) when compared to healthy young subjects and were similar to those of late-stage AD patients previously reported (1 = 0.94 0.03; 2 = 0.69 0.03; both p values> 0.1 [t-test])9,30. Thus, perhaps activity correlations at large time scales already had maximally degraded during earlier stages of dementia and AD due to dysfunction of the circadian pacemaker15. Alternatively, perhaps the follow-up duration was too short (only 83 individuals remained in the study aer 12 months). We favor the rst hypothesis because, while activity correlations at small time scales progressively decreased over the follow-up periods, those at large time scales remained stable regardless of treatment. To denitively distinguish these two possible mechanisms, studies of older people at the early or preclinical stage of the diseases as well as a longer follow-up period are warranted.

As a rst pilot investigation of within-subject changes in multiscale activity regulation, there are many notable limitations in the current study. One major potential concern is regarding the observed association between changes of activity correlations and cognition (MMSE). Since depression impairs both cognition35 and motor activity5,1114, is it possible that the association was a simple indirect inuence of worsening of depression with progression of dementia in our study? We have performed two additional statistical analyses to examine (1) the

SCIENTIFIC REPORTS

5

www.nature.com/scientificreports/

Figure 5. Association between functional outcomes and activity correlations at small time scales at baseline. (A) Mini-mental state examination (MMSE). (B) A subscale of the multidimensional observation scale for elderly subjects (MOSES) for withdrawn behavior. (C) Cornell Scale for Depression in Dementia (CSDD). (D) Negative aect scale of the Philadelphia Geriatric Centre Aect Rating Scale (PGCARS negative). Solid lines represent the predicted mean values and dashed lines are the 95% condence intervals.

eect of changes in activity correlations on MMSE changes aer controlling the eect of changes in Cornell Scale for Depression in Dementia (CSDD); and (2) the eect of changes in activity correlations on CSDD aer controlling the eect of changes in MMSE, respectively. We found that the association between changes in activity correlations and MMSE became not signicant (p> 0.1) while the association between changes in activity correlations and CSDD became less but still signicant (r = 0.13; t ratio = 2.91, p= 0.0039; before accounting for the eect of MMSE: r= 0.19, t ratio = 4.31, p < 0.0001). These results suggest that the observed association between degraded activity correlations and cognitive decline might be due to the inuences of depression on cognition and temporal activity regulation. However, due to the high correlation between the changes of MMSE and CSDD over time (r=0.80, t ratio=5.67; p< 0.0001) in the studied sample, the datasets are not ideal to tease out the complex causal links between depression, cognition, and multiscale activity regulation. In addition to depression, other neuropsychiatric symptoms such as hallucinations/delusions are also prevalent in dementia36.

How these dierent symptoms contribute to perturbed temporal activity patterns was not examined in the current study and is worth further investigation. Moreover, many other pathological factors (dierent from cognition and mood) such as physical disabilities, cardiac disorders, and metabolic diseases may aect mobility and, thus, can have independent inuences on activity correlations. We could not examine the potential role of these factors in degradation of activity regulation and its association with cognition in this study because no in-depth formal health examination/evaluation was made during the assessments that occurred every six months.

The other concern is that temporal activity correlations could only account for low percentages of variations in cognition and depression symptoms in patients/individuals with dementia, i.e., weak associations. Data integrity and study design might contribute to the weak associations. For instance, the activity recordings and the clinical measures were not collected simultaneously; and cognition and mood may vary signicantly from time to time (e.g., time of day, and time of year) while these outcome measures were not obtained at the same time for each assessment during the longitudinal study. Though activity correlations appear to be independent of the mean activity levels in healthy young individuals (e.g., correlations remain the same at home at dierent days and in the laboratory with restricted daily schedule and physical activity)6, no studies have formally tested the eects of

SCIENTIFIC REPORTS

6

www.nature.com/scientificreports/

Figure 6. Association between the changes of functional outcomes and changes in activity correlations at small time scales from baseline to the follow-up assessments. (A) Mini-mental state examination (MMSE). (B) A subscale of the multidimensional observation scale for elderly subjects (MOSES) for withdrawn behavior. (C) Cornell Scale for Depression in Dementia (CSDD). (D) Negative aect scale of the Philadelphia Geriatric Centre Aect Rating Scale (PGCARS negative). P values were obtained from mixed models in which all raw data were included (i.e., not binned categorized variable) with subject as a random factor for the intercept. Solid lines represent the predicted mean values and dashed lines are the 95% condence intervals.

daily schedule and home environments in older adults, especially in dementia patients whose daily activities rely much on their caregivers. In addition, multiscale activity regulation may also be temporarily altered by certain conditions such as shi work via their inuences on circadian control34. In this study, we calculated the correlations for each subject during each assessment using the whole recording while ignoring the potential variations in the correlations within the period of 12 weeks. Thus, the combination of non-simultaneous assessments and variations in activity control and clinical outcomes and might contribute to their weak associations.

Furthermore, the current study is focused on patients with dementia. Many neurodegenerative disorders can contribute to dementia such as AD, frontotemporal disorders, and Lewy body dementia. The progression of dementia and its impacts on behavior may dier for dierent types of dementia. The current study reveals the longitudinal changes of temporal structure in daily activity patterns in a heterogeneous population without distinguishing specic types of dementia. Future studies are warranted to examine whether within-subject changes of temporal activity correlations, their associations with cognition and mood, and their response to light are dierent in dierent types of dementia. In addition, within-subject changes of activity regulation in older persons without dementia are yet to be determined.

To test our hypotheses, we analyzed activity recordings collected from an existing database of 189 residents31. These residents (170 women and 19 men; 7096 years old; mean [SD] age: 85.7 [5.6] years) were living in assisted care facilities located in 12 dierent Dutch homes for the older individuals. The lowest age was data-driven rather than based on any cut-o exclusion criterion. These residents had their own apartment where they slept and retreated, but spent most of the daytime in a common living room supervised by caregivers. Among the 189 residents, there were 121 with probable AD based on the NINCDS-ADRDA criteria, 20 with probable vascular dementia and 24 with probable other types of dementia, based on the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) criteria for dementia and dementia subtypes37, 17 without dementia, and 7

SCIENTIFIC REPORTS

7

www.nature.com/scientificreports/

without sufficient information for reliable clinical diagnosis. The baseline demographic and clinical characteristics of subjects have been previously published31.

In a 2 2 factorial design, the 12 homes for the older individuals were randomly assigned to active (6 facilities, n= 98) or placebo (6 facilities, n= 91) light exposure, and subjects to double-blind daily intake of bedtime melatonin (n= 95) or placebo (n= 94). Thus, there are four groups: Group 1: 49 subjects assigned to active light and a placebo tablet; Group 2: 46 to inactive light and melatonin; Group 3: 49 to both active light and melatonin, and Group 4: 45 to inactive light and a placebo tablet. In this double blind randomized controlled trial, the four groups had no signicant dierences in age, sex, use of medication at inclusion and at any follow-up, and vision complications such as lens opacity and glaucoma31. The goal of this study is to determine the changes in temporal activity regulation and the clinical relevance in dementia patients. Thus, we studied all patients with conrmed dementia in the database (n = 165; 7096 years old), including 45 patients in Group 1, 39 in Group 2, took bedtime melatonin each day, 44 in Group 3, and 37 in Group 4. These patients were diagnosed with dementia primarily, not with Major Depressive Disorder although some were treated because of mood symptoms. The 4 groups did not dier with respect to the proportion of participants receiving psychotropic medication including antidepressants, antipsychotics, anxiolytics and hypnotics at the onset of their participation (2-tests, all p>0.32). Logistic mixed-eect regression analysis showed that the use of prescription drugs did not change aer treatment onset compared with prescription use prior to treatment onset (all p> 0.80). There were no eects on prescription with either light or melatonin treatment or their interaction (all p>0.35).

Informed consent was obtained from all subjects responsible relatives. The Medical Ethics Committees of Hospital De Gelderse Vallei, Ede, and the VU University Medical Center, Amsterdam, the Netherlands, approved the study. The methods were carried out in accordance with the approved guidelines.

Participants were followed for up to 3.5 years, with a mean (SD) of 15 (12) months. Recruitment and enrollment commenced in 1999 and data acquisition continued until 2004. Follow-up assessments were made 6 weeks aer the start of the treatment, and subsequently every 6 months. For each assessment, activity recordings and functional outcomes were assessed as described below.

Melatonin (2.5 mg, Terafarm, Brielle, the Netherlands) was given to Groups 2 and 3 approximately 1 hour before bedtime by the nursing sta who ensured adherence. The tablets took about 1hour to completely dissolve in water (a medium-fast release preparation). Timing and dosage were based on previous studies38,39. Placebo was given to Groups 1 and 4 according to the same schedule as for the other two groups.

Light exposure was manipulated by installing a large number of ceiling-mounted xtures with Plexiglas diffusers containing an equal amount of Philips TLD 840 and 940 uorescent tubes (Philips Lighting BV, Eindhoven, the Netherlands) in the common living room of each of the selected 6 facilities. Illumination levels were obtained at intervals throughout one day at each assessment using a lux meter held at eye level in the direction of gaze, which was usually slightly downward or at best representing light falling on the vertical plane. A total of 3017 assessments were made. Average light exposure measured at eye level in the gaze direction was increased to ~1000 lux between 10 AM and 6 PM at the facilities randomized to the active light condition (P < 0.01 for all hourly comparisons between 10 AM and 6 PM of the active condition versus baseline except between 3 and 4 PM). The intensity for the active light treatment condition is enough to synchronize circadian rhythms in healthy people in a time-free environment40 and to improve daily/circadian activity rhythm disturbances in older patients with dementia41. For the placebo group in the other 6 facilities, an equal number of xtures with only half of the tubes and concealed band-stop lters were installed at a greater distance from the eyes to achieve an exposure of ~300 lux. This clinical trial was registered in controlled-trials.com on December 9, 2005 (isrctnIdentier:ISRCTN93133646). Clinical outcomes from this trial of bright light and melatonin in dementia have previously been published31.

For each assessment that was performed at baseline, aer 6 weeks of treatment onset, and subsequently every 6 months aer treatment onsets, motor activity levels were continuously monitored for 12 weeks using an Actiwatch (Cambridge Neurotechnology, Cambridge, England) worn on the wrist of the non-dominant hand. Acceleration was sampled at 32Hz and was integrated to a proprietary count value every minute.

The primary clinical outcomes are cognitive function assessed with the minimental state examination (MMSE)42, and mood with the Cornell Scale for Depression in Dementia (CSDD)43,44. In addition, we considered two secondary outcome measures: the Philadelphia Geriatric Centre Aect Rating Scale (PGCARS)45 and the withdrawn behavior subscale of the multidimensional observation scale for elderly subjects (MOSES)46.

To assess activity regulation at multiple time scales, we performed detrended uctuation analysis (DFA) to examine temporal correlations in the activity uctuations at time scales from ~0.1 up to 12hours. This method quanties the detrended uctuation function, F(n), of activity uctuations at dierent time scales n (Fig.1)47. To eliminate the eect of possible linear trends in original data, we applied the 2nd order DFA, i.e., the 2nd order of polynomial functions were used to detrend data when calculating F(n)48. A power-law form of F(n) indicates self-similarity (scale-invariance) in the uctuations, yielding F(n)~n. The parameter , called the scaling exponent, quanties the correlation property in the signal as follows: if = 0.5, there are no correlations in the uctuations (white noise); if > 0.5, there are positive correlations, where large activity values are more likely to be followed by large activity values (and vice versa for small activity values). The exponent = 1.0 indicates highest complexity in the systems33,49. Similar values close to 1.0 have been observed in many physiological outputs under normal conditions, indicating a biological system with complex temporal regulations33.

SCIENTIFIC REPORTS

8

www.nature.com/scientificreports/

In this study, we focused on multiscale activity regulation at time scales 12hours (instead of up to 24hours) during daytime. We made the decision for a number of reasons. (1) Scheduled 24-hour events, as likely occurred in assisted care facilities, can signicantly aect activity uctuations at time scales close to 24hours (i.e., a masking eect)9. As previously reported, activity correlation and its changes with aging and in dementia can be properly identied using only daytime data6,9. (2) During the scheduled sleep episodes, activity of participants was somehow restricted because they stayed in their own apartment without access to other facilities. To avoid the potential eects of imposed daily scheduled events, data during the individually assessed scheduled bedtimes (nurse informants) were excluded for analysis. Consequently, the derived activity measures reect activity regulation that is more independent of sleep dynamics. Previous cross-sectional studies suggested that scale-invariant correlations are disrupted with aging and in dementia, leading to distinct correlations over two time scale regions with the boundary at ~1.52 hours9,10. Thus, for each activity recording in this study, we calculated the scaling exponent in two regions, separately, i.e. 1 at <90minutes and 2 at >2hours, omitting the variable transitional region of time scales between 1.52hours.

Activity recordings of 21 dementia patients were either missing or were too short to be used for the correlation analysis. Thus, DFA results of 144 dementia subjects were reported in this study, including 40 (29 probable AD) in Group 1, 34 (25 probable AD) in Group 2, 39 (29 probable AD) in Group 3, and 30 (18 probable AD) in Group 4.

Follow-up time, temporal correlations of activity uctuations, and measures of cognitive performance and mood were evaluated as continuous variables. Statistical analyses were performed using JMP Pro 11 (SAS Institute, Cary, NC). ANOVAs were used to determine the group dierences in temporal correlations of activity uctuations and measures of cognitive performance and mood at baseline. Mixed models with subject as a random factor for intercept were used to determine the eects of follow-up time, treatment with light, melatonin and their interaction on temporal correlations of activity uctuations. Mixed models subsequently assessed the associations of changes in temporal correlations of activity uctuations with changes in the measures of cognitive performance and mood that occurred from baseline to the nal follow-up assessment.

References

1. Alzheimers Association. 2015 Alzheimers disease facts and gures. Alzheimers Dement. J. Alzheimers Assoc. 11, 332384 (2015).2. Ortman, J. M., Velko, V. A. & Hogan, H. An Aging Nation: The Older Population in the United States. U.S. Department of Commerce. Economics and Statistics Administration. U.S. Census Bureau (2014) (Date of access: 11/25/2015). https://www.census.gov/ prod/2014pubs/p25-1140.pdf.

3. Hayes, T. L. et al. Unobtrusive assessment of activity patterns associated with mild cognitive impairment. Alzheimers Dement. J. Alzheimers Assoc. 4, 395405 (2008).

4. Aybek, S. et al. Fractal temporal organisation of motricity is altered in major depression. Psychiatry Res. 200, 288293 (2012).5. Indic, P. et al. Multi-scale motility amplitude associated with suicidal thoughts in major depression. PloS One 7, e38761 (2012).6. Hu, K. et al. Non-random uctuations and multi-scale dynamics regulation of human activity. Phys. -Stat. Mech. Its Appl. 337, 307318 (2004).

7. Amaral, L. a. N. et al. Power law temporal auto-correlations in day-long records of human physical activity and their alteration with disease. Europhys. Lett. 66, 448454 (2004).

8. Ohashi, K. et al. Decreased fractal correlation in diurnal physical activity in chronic fatigue syndrome. Methods Inf. Med. 43, 2629 (2004).

9. Hu, K., Van Someren, E. J. W., Shea, S. A. & Scheer, F. A. J. L. Reduction of scale invariance of activity uctuations with aging and Alzheimers disease: Involvement of the circadian pacemaker. Proc. Natl. Acad. Sci. USA 106, 24902494 (2009).

10. Hu, K., Harper, D. G., Shea, S. A., Stopa, E. G. & Scheer, F. A. J. L. Noninvasive fractal biomarker of clock neurotransmitter disturbance in humans with dementia. Sci. Rep. 3, 2229 (2013).

11. Indic, P. et al. Scaling behavior of human locomotor activity amplitude: association with bipolar disorder. PloS One 6, e20650 (2011).12. Krane-Gartiser, K., Henriksen, T. E. G., Morken, G., Vaaler, A. & Fasmer, O. B. Actigraphic assessment of motor activity in acutely admitted inpatients with bipolar disorder. PloS One 9, e89574 (2014).

13. Sano, W. et al. Enhanced persistency of resting and active periods of locomotor activity in schizophrenia. PloS One 7, e43539 (2012).14. Nakamura, T. et al. Of mice and menuniversality and breakdown of behavioral organization. PloS One 3, e2050 (2008).15. Pittman-Polletta, B. R., Scheer, F. A. J. L., Butler, M. P., Shea, S. A. & Hu, K. The role of the circadian system in fractal neurophysiological control. Biol. Rev. 88, 873894 (2013).

16. Lim, A. S. P. & Saper, C. B. Sleep, circadian rhythms, and dementia. Ann. Neurol. 70, 677679 (2011).17. Monteleone, P. & Maj, M. The circadian basis of mood disorders: recent developments and treatment implications. Eur. Neuropsychopharmacol. J. Eur. Coll. Neuropsychopharmacol. 18, 701711 (2008).

18. Albrecht, U. Circadian clocks and mood-related behaviors. Handb. Exp. Pharmacol. 227239, doi: 10.1007/978-3-642-25950-0_9 (2013).

19. Salvatore, P., Indic, P., Murray, G. & Baldessarini, R. J. Biological rhythms and mood disorders. Dialogues Clin. Neurosci. 14, 369379 (2012).

20. Tranah, G. J. et al. Circadian activity rhythms and risk of incident dementia and mild cognitive impairment in older women. Ann. Neurol. 70, 722732 (2011).

21. Ju, Y.-E. S. et al. Sleep quality and preclinical Alzheimer disease. JAMA Neurol. 70, 587593 (2013).22. Lim, A. S. P., Kowgier, M., Yu, L., Buchman, A. S. & Bennett, D. A. Sleep Fragmentation and the Risk of Incident Alzheimers Disease and Cognitive Decline in Older Persons. Sleep 36, 10271032 (2013).

23. Van Cauter, E. & Turek, F. W. Depression: a disorder of timekeeping? Perspect. Biol. Med. 29, 510519 (1986).24. Soutre, E. et al. Circadian rhythms in depression and recovery: evidence for blunted amplitude as the main chronobiological abnormality. Psychiatry Res. 28, 263278 (1989).

25. Li, J. Z. et al. Circadian patterns of gene expression in the human brain and disruption in major depressive disorder. Proc. Natl. Acad. Sci. USA 110, 99509955 (2013).

26. Forbes, D., Blake, C. M., Thiessen, E. J., Peacock, S. & Hawranik, P. Light therapy for improving cognition, activities of daily living, sleep, challenging behaviour, and psychiatric disturbances in dementia. Cochrane Database Syst. Rev. 2, CD003946 (2014).

27. Alzoubi, K. H. et al. Chronic Melatonin Treatment Prevents Memory Impairment Induced by Chronic Sleep Deprivation. Mol. Neurobiol., doi: 10.1007/s12035-015-9286-z (2015).

28. Swaab, D. F., Van Someren, E. J., Zhou, J. N. & Hofman, M. A. Biological rhythms in the human life cycle and their relationship to functional changes in the suprachiasmatic nucleus. Prog. Brain Res. 111, 349368 (1996).

SCIENTIFIC REPORTS

9

www.nature.com/scientificreports/

29. Wang, J. L. et al. Suprachiasmatic neuron numbers and rest-activity circadian rhythms in older humans. Ann. Neurol. 78, 317322 (2015).

30. Hu, K., Scheer, F. A. J. L., Ivanov, P. C., Buijs, R. M. & Shea, S. A. The suprachiasmatic nucleus functions beyond circadian rhythm generation. Neuroscience 149, 508517 (2007).

31. Riemersma-van der Lek, R. F. et al. Eect of bright light and melatonin on cognitive and noncognitive function in elderly residents of group care facilities: a randomized controlled trial. JAMA 299, 26422655 (2008).

32. Robinson, L., Tang, E. & Taylor, J.-P. Dementia: timely diagnosis and early intervention. BMJ 350, h3029 (2015).33. Goldberger, A. L. et al. Fractal dynamics in physiology: alterations with disease and aging. Proc. Natl. Acad. Sci. USA 99 Suppl 1, 24662472 (2002).

34. Hsieh, W.-H. et al. Simulated shi work in rats perturbs multiscale regulation of locomotor activity. J. R. Soc. Interface 11, 20140318 (2014).

35. Bennett, S. & Thomas, A. J. Depression and dementia: cause, consequence or coincidence? Maturitas 79, 184190 (2014).36. Borsje, P., Wetzels, R. B., Lucassen, P. L., Pot, A. M. & Koopmans, R. T. The course of neuropsychiatric symptoms in community-dwelling patients with dementia: a systematic review. Int. Psychogeriatr. IPA 27, 385405 (2015).

37. Association, A. P. Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, Text Revision. (American Psychiatric Association, 2000).

38. Zhdanova, I. V., Lynch, H. J. & Wurtman, R. J. Melatonin: a sleep-promoting hormone. Sleep 20, 899907 (1997).39. Singer, C. et al. A multicenter, placebo-controlled trial of melatonin for sleep disturbance in Alzheimers disease. Sleep 26, 893901 (2003).

40. Middleton, B., Stone, B. M. & Arendt, J. Human circadian phase in 12:12h, 200: <8 lux and 1000: <8 lux light-dark cycles, without scheduled sleep or activity. Neurosci. Lett. 329, 4144 (2002).

41. Van Someren, E. J., Kessler, A., Mirmiran, M. & Swaab, D. F. Indirect bright light improves circadian rest-activity rhythm disturbances in demented patients. Biol. Psychiatry 41, 955963 (1997).

42. Folstein, M. F., Folstein, S. E. & McHugh, P. R. Mini-mental state. A practical method for grading the cognitive state of patients for the clinician. J. Psychiatr. Res. 12, 189198 (1975).

43. Alexopoulos, G. S., Abrams, R. C., Young, R. C. & Shamoian, C. A. Cornell Scale for Depression in Dementia. Biol. Psychiatry 23, 271284 (1988).

44. Harwood, D. G., Ownby, R. L., Barker, W. W. & Duara, R. The factor structure of the Cornell Scale for Depression in Dementia among probable Alzheimers disease patients. Am. J. Geriatr. Psychiatry O. J. Am. Assoc. Geriatr. Psychiatry 6, 212220 (1998).45. Lawton, M. P., Van Haitsma, K. & Klapper, J. Observed aect in nursing home residents with Alzheimers disease. J. Gerontol. B. Psychol. Sci. Soc. Sci. 51, P314 (1996).

46. Helmes, E., Csapo, K. G. & Short, J. A. Standardization and validation of the Multidimensional Observation Scale for Elderly Subjects (MOSES). J. Gerontol. 42, 395405 (1987).

47. Peng, C. K. et al. Fractal mechanisms and heart rate dynamics. Long-range correlations and their breakdown with disease. J. Electrocardiol. 28 Suppl, 5965 (1995).

48. Hu, K., Ivanov, P. C., Chen, Z., Carpena, P. & Stanley, H. E. Eect of trends on detrended uctuation analysis. Phys. Rev. E 64, 11114 (2001).

49. Lo, M. T. et al. Interactive Eects of Dorsomedial Hypothalamic Nucleus and Time-Restricted Feeding on Fractal Motor Activity Regulation. Front. Physiol. 7, 174 (2016).

We thank Philips Lighting BV and Cambridge Neurotechnology for supplied material at reduced cost and the research volunteers for their participation. We thank Dr. C. Saper for his insightful comments on the pathological mechanisms underlying the degraded temporal activity regulation in dementia. We also thank two reviewers for their constructive suggestions, which improved the presentation and interpretation of our results. This work was supported by NIH grants R00-HL102241, R01AG048108-01A1 and P01AG009975 to K.H.; K24-HL076446 to S.A.S.; R01-HL118601 to F.A.J.L.S.; the China Postdoctoral Council, the International Postdoctoral Exchange Fellowship 20150042 to P.L.; the Netherlands Organization for Health Research, the Hague, grants 0028-300-30 and 907-00-012, the Netherlands Organisation for Scientic Research, the Hague, grant VICI 453-07-001, Hersenstichting Nederland grant 11F04-2.47, and Internationale Stichting Alzheimer Onderzoek grant 05511 to E.J.W.V.S.

Author Contributions

K.H., F.A.J.L.S., S.A.S. and E.J.W.V.S. designed the study; R.F.R.-v.d.L. and E.J.W.V.S. collected the data; K.H., M.P., P.L. and E.J.W.V.S. analyzed the data and performed statistical analyses; K.H. and E.J.W.V.S. draed the manuscript; and all authors contributed to and approved the nal version.

Additional Information

Competing nancial interests: The authors declare no competing nancial interests.

How to cite this article: Hu, K. et al. Progression of Dementia Assessed by Temporal Correlations of Physical Activity: Results From a 3.5-Year, Longitudinal Randomized Controlled Trial. Sci. Rep. 6, 27742; doi: 10.1038/ srep27742 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

SCIENTIFIC REPORTS

10