Relationship between physical activity and biomarkers of pathology and neuroinflammation in preclinical autosomal‐dominant Alzheimer's disease

Abstract

Objective

Physical activity (PA) has been linked to reduced Alzheimer's disease (AD) risk. However, less is known about its effects in the AD preclinical stage. We aimed to investigate whether greater PA was associated with lower plasma biomarkers of AD pathology, neural injury, reactive astrocytes, and better cognition in individuals with autosomal‐dominant AD due to the presenilin‐1 E280A mutation who are virtually guaranteed to develop dementia.

Methods

Twenty‐eight cognitively unimpaired mutation carriers (ages x̄ = 29.28) wore a FitBit Charge‐4 for 14 days. We calculated their average steps to measure locomotion, and Training Impulse (TRIMP) to quantify the intensity and duration of PAs using heart rate. Plasma amyloid beta 42/40 ratio, phosphorylated tau 181, neurofilament light chain, and glial fibrillary acidic protein (GFAP) were measured. Cognition was assessed with the Consortium to Establish a Registry for Alzheimer's Disease word list learning and delayed recall, Trail Making Test Part A, and Wechsler Adult Intelligence Scale‐version IV Digit Span Backward. We conducted multiple linear regressions controlling for age, sex, body mass index, and education.

Results

There were no associations among steps or TRIMP with plasma biomarkers or cognition. Greater TRIMP was related to higher GFAP levels.

Conclusions

PA was not associated with cognition or plasma biomarkers. However, greater intensity and duration of PAs were related to higher GFAP. Participants engaged very little in moderate to vigorous PA. Therefore, light PA may not exert a significant protective effect in preclinical AD. Future work with larger samples and longitudinal data is needed to elucidate further the potential impact of PA on AD progression in the preclinical stages.

Highlights

Locomotion (average steps) was not associated with plasma biomarkers or cognition. Greater training load (training impulse) was related to higher glial fibrillary acidic protein levels in mutation carriers. Light physical activity may not suffice to exert a protective effect on Alzheimer's disease.

INTRODUCTION

Alzheimer's disease (AD) is a significant public health concern that is expected to become more prevalent as the population ages in the absence of a cure.¹ AD is characterized by the progressive accumulation of amyloid plaques and hyperphosphorylated tau protein that leads to synaptic dysfunction and neurodegeneration, and eventually to dementia.^2,3 Amyloid beta (Aβ) accumulates in the cortex nearly 2 decades before the onset of cognitive impairment (i.e., the preclinical stage), followed by hyperphosphorylated tau.^3–5 Increasing evidence shows that dysfunction of glial cells further contributes to pathology accumulation, neurodegeneration, and cognitive decline.^6,7 It has been suggested that interventions targeting these processes, whether pharmacological or behavioral, hold promise for potentially slowing the progression of the disease, thereby delaying the onset of impairment.

Physical activity (PA; i.e., any bodily movement)⁸ and exercise (i.e., a PA that is planned, structured, and repetitive)⁸ have been postulated as promising lifestyle factors that could protect against dementia.⁹ Greater PA and exercise have been associated with greater brain volume and function, and better cognitive performance across the lifespan.^10–13 Exercise-randomized clinical trials with healthy older adults have shown significant improvements in cognitive functioning, increased hippocampal volume and thickness, and stronger functional brain networks.^11,12,14 Further, some studies with people who are at risk for developing dementia have also shown that greater self-reported PA is related to less neurodegeneration and cognitive decline.^15–18 For instance, a study with healthy older adults who were amyloid positron emission tomography (PET) positive and engaged in greater PA, as measured by a pedometer, exhibited significantly slower cognitive decline and gray matter volume loss over 6 years compared to those who engaged in low PA.¹⁹ Similarly, in a study with a large cohort of late-middle-aged adults at risk for AD, findings showed that individuals with high self-reported PA levels exhibited lower age-related alterations in PET Aβ deposition, glucose metabolism, hippocampal volume, and cognitive functioning compared to those who were physically inactive.²⁰ Further, research in autosomal-dominant AD (ADAD) has also suggested that PA may influence AD progression, as one study showed that carriers of mutations that cause early-onset AD and who self-reported engaging in high levels of PA experienced a slower increase in cerebrospinal fluid biomarkers of tau and Aβ with respect to estimated years from the onset of cognitive impairment, compared to mutation carriers with low PA levels.²¹

However, little is still known about how PA impacts the progression of AD in humans who are in the preclinical stage of AD. Understanding whether and how PA may modify AD progression or cognitive function in the preclinical stage of AD will be valuable in informing interventions for those who are in the early AD stages. Several barriers have made it very difficult to closely examine these phenomena in older adults: (1) the inability to predict who will develop AD and dementia and (2) the presence of confounding factors that make it almost impossible to disentangle the contribution of PA to AD pathology versus cardiovascular risk, among others. Further, many studies, including in ADAD, have measured self-reported PA, which is vulnerable to recall bias.²²

The current study aimed to examine the association of PA with plasma biomarkers of AD pathology, neural injury, astrocyte reactivity, and cognitive functioning in cognitively unimpaired presenilin-1 (PSEN1) E280A mutation carriers with ADAD. PSEN1 E280A carriers will invariably develop dementia due to AD in their late 40s and have a well-characterized clinical profile.²³ Further, PSEN1 E280A mutation carriers have minimal confounding factors that could impact the brain and cognitive functioning (e.g., cardiovascular risk factors, older age). As such, studying PSEN1 E280A mutation carriers presents a unique opportunity for understanding the relationship of PA with AD progression during the preclinical stage as they already exhibit evidence of AD pathology and are otherwise healthy.

RESEARCH IN CONTEXT

• Systematic Review: We reviewed the literature on physical activity (PA), exercise, plasma biomarkers, and cognition in Alzheimer's disease (AD) using traditional sources (e.g., PubMed). Our search revealed that little is known about the relationship between PA and AD biomarkers in the preclinical stage.

• Interpretation: We reported that PA, both locomotion and training load, was not associated with AD plasma biomarkers or cognition in non-demented presenilin-1 E280A carriers with autosomal-dominant AD who mostly engaged in light PA, except for a marker of astrocyte reactivity. Thus, engaging in higher intensity PA may be necessary to confer significant protection in AD preclinical stages.

• Future Directions: Findings set the stage for studies to investigate the relationship between PA and plasma biomarkers of AD in people who engage in greater intensity PA, as well as how PA predicts change in plasma biomarkers and cognitive decline using longitudinal data.

We measured two aspects of PA that have been associated with health outcomes using a wearable device: locomotion and training load.^24–26 Locomotion, referring to physical movement regardless of intensity, was measured by quantifying the number of steps.²⁷ Training load, a comprehensive measure of the overall effort exerted during PA, was assessed using a modified training impulse (TRIMP).^28,29 TRIMP accounts for the intensity and duration of PA based on heart rate (HR) data.^28–30 Each of these measures provides valuable insight into aspects of PA that may influence brain and cognitive functioning: One evaluates whether overall movement may potentially exert a positive effect on AD plasma biomarkers, while the other examines if the intensity and duration of the PA are key for brain protection.

Further, recent advances in proteomics have enabled the measurement of AD pathological processes in plasma. Substantial evidence shows that a lower Aβ42/40 ratio and elevated phosphorylated tau 181 (p-tau181) are strongly associated with amyloid plaques and neurofibrillary tangles, respectively, and are strong predictors of cognitive decline.^31–33 Neurofilament light chain (NfL), a biomarker of neural injury, is also elevated before symptoms emerge and associated with brain pathology, neurodegeneration, and cognitive functioning.^31,34 More recently, glial fibrillary acidic protein (GFAP) has garnered attention as a biomarker of reactive astrocytes and a strong predictor of dementia risk.^31,35 These biomarkers also distinguish PSEN1 E280A carriers from non-carriers several years before estimated symptom onset and are associated with brain pathology in this population.^34,36,37 Therefore, these plasma biomarkers can be considered a reflection of brain changes associated with AD. In this context, we tested the hypothesis that greater locomotion and training load would be associated with higher plasma levels of Aβ42/40, and lower plasma levels of p-tau181, NfL, and GFAP in cognitively unimpaired PSEN1 E280A carriers.

METHODS AND MATERIALS

Study design and participants

A total of 28 PSEN1 E280A mutation carriers from the same kindred who are enrolled in the Massachusetts General Hospital (MGH) Colombia-Boston (COLBOS) biomarker study or the COLBOS Remote study, which was initiated during the pandemic, participated. All participants were recruited from the Alzheimer's Prevention Initiative registry of familial AD, which includes > 6000 living members of the kindred and ≈ 1200 mutation carriers.²² PSEN1 E280A carriers usually show symptoms of mild cognitive impairment (MCI) at a median age of 44 years and dementia at 49 years, but measurable changes in memory appear nearly 12 years before MCI.²³ To be included in this study, participants had to demonstrate no cognitive impairment ( ≤ 1.5 standard deviation [SD]) on a standard cognitive battery, a score of 0 on a clinical diagnostic rating scale (Clinical Dementia Rating [CDR]), a Functional Assessment Staging Test (FAST) score of ≤ 2, and a Mini-Mental State Examination (MMSE) score of ≥ 26. Exclusion criteria included a history of psychiatric disorders, illiteracy, stroke, epilepsy, traumatic brain injury, kidney failure, human immunodeficiency syndrome, or substance abuse. Demographic information is presented in Table 1.

TABLE 1 Demographic, clinical, and plasma biomarkers data.

	Mean (SEM)
n	28
Age (years)	29.14 (0.64)
Formal education (years)	12.68 (0.46)
Sex (% female)	68
BMI	23.03 (0.47)
MMSE	28.18 (0.24)
FAST (stages 1,2)	19, 9
Average steps	10,292.56 (996.79)
TRIMP	2060.64 (246.98)
Light (min at 40% to 59% of maximum HR)	1953.26 (241.24)
Moderate (min at 60% to 79% of maximum HR)	98.49 (20.10)
Vigorous (min at >80% of maximum HR)	8.89 (5.43)
Amyloid-β 42/40 (pg/mL)	0.08 (0.002)
ptau-181 (pg/mL)	19.84 (1.56)
Neurofilament light (pg/mL)	7.08 (0.69)
Glial fibrillary acidic protein (pg/mL)	70.76 (6.52)
CERAD word list learning	20.75 (0.75)
CERAD word list delayed recall	7.68 (0.29)
TMT-A (time in seconds)	40.21 (2.64)
WAIS-IV Digit Span Backward	5.46 (0.39)

Consent statement

The study was approved by both the institutional ethics review boards of the University of Antioquia in Medellín, Colombia, and the Mass General Brigham in Boston, Massachusetts. All participants provided written informed consent before participating in any procedures.

Clinical and cognitive assessments

Cognitive tests were performed at the University of Antioquia in Medellín, Colombia in Spanish by neuropsychologists or by psychologists trained in neuropsychological assessment. Participants underwent a clinical interview and were administered the MMSE, and the FAST, a measure that assesses functional status.³⁸ A person in stage 1 is cognitively normal with no subjective cognitive decline, and stage 2 refers to someone who is cognitively normal with subjective cognitive decline.

Participants were also administered the Spanish version of the Consortium to Establish a Registry for Alzheimer's Disease (CERAD) word list learning and delayed recall, which is sensitive to early memory changes in PSEN1 E280A carriers.²³ In this test, participants were asked to recall a list of 10 unrelated items that were repeated three times (i.e., learning), and to say as many words from the list as they could remember after a 10-minute delay (i.e., delayed recall). A total score for learning was calculated by adding the total number of words recalled correctly during the learning phase (/30), and a delayed recall score with the total number of words recalled after the delay (/10). In addition, participants completed the Trail Making Test Part A (TMT-A), a measure of psychomotor speed that requires participants to connect numbers in sequential order as fast as possible, and the Wechsler Adult Intelligence Scale-version IV (WAIS-IV) Digit Span Backward, a measure of verbal working memory that requires participants to repeat a string of numbers of increasing length backward.

Genotyping

Genomic DNA was extracted from the blood by standard protocols, and PSEN1 E280A characterization was done at the University of Antioquia using methods previously described.³⁹ Participants and investigators were blinded to the genetic status of the individual.

PA measures

PA was measured using a wrist-worn FitBit Charge 4. Participants were instructed to wear the FitBit on either wrist for at least 3 weeks in Colombia, engage in activities as usual, and take it off when showering or charging it while they engaged in a sedentary activity such as reading or watching TV. Here, we report data for 14 consecutive days during which participants wore their FitBit for > 85% of the day.⁴⁰ We report two variables: (1) average steps and (2) TRIMP. Average steps were calculated by averaging the total number of steps per day for 14 consecutive days. As such, the average steps variable represents the average group number of daily steps taken over 14 days. A modified TRIMP was calculated using their daily HR data, which is collected ≈ every 5 seconds. First, we used the daily average resting HR in beats per min (bpm) during sleep that FitBit provides to calculate the 14-day average resting HR.⁴¹ We then used this value to calculate HR reserve (HRR) by subtracting the average resting HR from the individual's estimated maximum HR, following Fox and Haskell's formula (i.e., 220 – age). We calculated TRIMP by multiplying the training intensity, measured using bpm, by the training volume, measured in minutes. Specifically, we used the following formula to determine the bpm range that represented 40% to 59% of the maximum HR for each individual (light PA): ([HHR x 0.4] + average resting HR). The HHR was multiplied by 0.6 to determine the bpm range that represented 60% to 79% of their maximum HR (moderate activity) and by 0.8 to determine the bpm range that represented > 80% of their maximum HR (vigorous activity). These values were then each multiplied by the time spent in that HR zone for each person (e.g., ([HHR x 0.4] + average resting HR) x minutes spent in that HR zone). The products were added to yield a TRIMP value.

Biochemical measures and analysis

All plasma samples were collected at the University of Antioquia in Medellín, Colombia. Plasma was collected in the morning (non-fasting collection). Samples were stored at −80°C. Analytes reported here were measured using bead-based single molecule array (SIMOA) assays and performed using a fully automated Quanterix HD-X analyzer (Quanterix). Assays were performed according to the manufacturer's instructions at the MIND Biomarker Core after determining optimal dilution factors by serially diluting four pooled plasma samples within the range of published recommendations. All samples were assayed in duplicate and serially diluted standard curves were included on every plate.

The sensitivity of each assay was determined by comparing sample concentrations to the assay's lower limit of quantification (LLOQ) defined as the concentration of the lowest calibrator with a coefficient of variation (CV) < 20% and a recovery between 80% and 120% of expected values. A biomarker assay was considered satisfactorily sensitive if the analyte was consistently measured above LLOQ in > 50% of samples using the experimentally determined dilution factor. Intra-assay reliability was determined by calculating the median CV for sample replicates. Analytes with median intra-assay replicate CVs < 20% were considered to have acceptable technical reliability. To evaluate short-term biotemporal stability, analyte concentrations were measured in the three plasma samples from each individual included on the same plate to minimize plate-to-plate variability. Sample concentrations were calculated against the standard curve using the SIMOA HD-X analyzer software (Quanterix).

Statistical analyses

We conducted descriptive statistics for all variables. We conducted separate linear regressions to examine the association of average steps and TRIMP with age, a proxy of disease progression in PSEN1 E280A carriers. We conducted the Shapiro–Wilk test to determine whether the dependent variables (i.e., biomarkers) were normally distributed. Based on the results, we rejected the null hypotheses for Aβ42/40 (P = 0.022), NfL (P = 0.003), and GFAP (P ≤ 0.001), suggesting that these data do not follow a normal distribution. As such, we log-transformed each biomarker variable. We then conducted multiple linear regressions controlling for age, sex assigned at birth, and body mass index (BMI) to test the association of average steps and TRIMP with plasma biomarkers and cognitive test scores. We also controlled for education in all models examining cognition. Of note, the relationship between PA (i.e., average steps and TRIMP) and cognitive test scores was evaluated in 30 mutation carriers with an average age of 29.57 years (SD: 4.04; range: 23–42 years) who had FitBit and cognitive data but not biomarkers. Data points with a SD of  ≥ 3 from the mean were considered outliers. However, there were no outliers and thus, no data were excluded from the analyses. Analyses used a family-wise significance threshold of P < 0.05 and were performed using SPSS (V.28.0; SPSS Inc.).

RESULTS

PA and plasma biomarkers

Descriptive statistics for PA are presented in Table 1. Average steps were not significantly associated with age (β = 0.28, P = 0.151). Similarly, average steps did not significantly predict levels of Aβ 42/40 (β = –0.19, P = 0.431), p-tau181 (β = 0.13, P = 0.575), NfL (β = 0.007, P = 0.975), or GFAP (β = 0.37, P = 0.080) when accounting for age, sex assigned at birth, and BMI (Figure 1).

[IMAGE OMITTED. SEE PDF]

TRIMP values were not significantly associated with age (β = –0.01, P = 0.965). They also did not significantly predict Aβ42/40 (β = –0.37, P = 0.153), p-tau181 (β = 0.35, P = 0.161), or NfL (β = 0.34, P = 0.160) levels (Figure 2). However, higher TRIMP levels predicted greater GFAP levels when controlling for age, sex assigned at birth, and BMI (β = 0.52, P < 0.019, 95% confidence interval: 0.000016, 0.000155; Figure 2). Because the dependent variable was log-transformed, we exponentiated the coefficient showing that for every one-unit increase in the TRIMP value, GFAP levels increase by a factor of ≈ 1.

[IMAGE OMITTED. SEE PDF]

PA and cognitive functioning

Average steps were not significantly associated with word list learning (β = 0.25, P = 0.238), delayed recall (β = 0.22, P = 0.296), TMT-A (β = 0.002, P = 0.991), or WAIS-IV Digit Span Backward (β = 0.09, P = 0.654) when accounting for age, sex assigned at birth, years of education, and BMI.

TRIMP values were also not significantly associated with word list learning (β = 0.17, P = 0.516), delayed recall (β = 0.36, P = 0.176), TMT-A (β = 0.10, P = 0.688), or WAIS-IV Digit Span Backward (β = 0.20, P = 0.447) when controlling for age, sex assigned at birth, years of education, and BMI scores.

DISCUSSION

PA has been identified as a lifestyle factor that may reduce the risk of brain disease and promote healthy aging.⁹ However, less is known about its potential effects on people who are in the preclinical stage of AD. Further, there is much discussion about what aspects of PA are essential for reducing the risk of AD and other brain diseases; that is, whether it is leading a life that is not sedentary and engaging in any PA or if most of these activities must be at least moderate in intensity relative to the person's age. We addressed these gaps by examining PA, both locomotion and training load, which considers the intensity and duration of PA, and plasma biomarkers of AD in young cognitively unimpaired individual PSEN1 E280A carriers with ADAD.

Our findings showed that mutation carriers took an average of ≈ 10,000 daily steps over 14 days and engaged in mostly light activity. There were no statistically significant associations between PA levels and age. As such, levels of PA did not vary based on proximity to the expected onset of MCI in this population. Average steps and training load were not statistically associated with plasma biomarkers of AD pathology and neural injury, or with cognitive functioning. However, greater training load was significantly associated with higher GFAP levels.

Contrary to what we expected, we did not find evidence of PA as a factor that could potentially offset the effects of AD pathology and related impact on cognition in the preclinical stages of AD. A study from the Dominantly Inherited Alzheimer's Network (DIAN) examined differences in cerebrospinal fluid biomarkers of amyloid plaques (Aβ1-42), total tau (t-tau), p-tau181, t-tau/Aβ1-42, p-tau/Aβ1-42, and brain amyloid burden between 224 mutation carriers who engaged in high versus low PA as measured by a self-reported questionnaire that assessed the average time spent in 10 activities over a year.²¹ They found that PA decreased over time in mutation carriers. Further, higher PA (> 150 minutes/week) was associated with higher MMSE scores and lower CDR scale Sum of Boxes (CDR-SOB) scores, a measure of cognitive and functioning performance, at baseline. Mutation carriers in the low PA group showed a greater decline in MMSE scores and an increase in CDR-SOB compared to the high PA group. Those in the low PA group exhibited a more pronounced increase in t-tau and tau/Aβ1-42 relative to the high PA group with respect to estimated years from the onset of cognitive impairment. Notably, 70% of mutation carriers fell in the high PA group. While we did not directly measure total time spent in PA of at least moderate intensity, we can infer from the HR data and TRIMP values that most of our participants spent less time in at least moderately intense PA. This likely contributed to the discrepancy in our findings because, as recommended by the American College of Sports Medicine,⁴² the beneficial effects of PA are mostly observed when people engage in moderate-intensity aerobic PA for a minimum of 30 minutes on 5 days per week, or vigorous-intensity aerobic activity for a minimum of 20 minutes on 3 days per week. Further, PA data from the DIAN cohort is based on self-report, which may also account for the differences between the cohorts in PA levels. Notably, some intervention studies with older adults at risk for dementia have not found associations between PA and AD biomarkers and cognition.^43,44 A recent intervention study testing the effect of a 6-month exercise intervention on several biomarkers, including Aβ42/40, p-tau181, NfL, and GFAP in cognitively unimpaired older adults, found no changes in biomarkers pre- and post-intervention.⁴⁵ More research is needed to elucidate if moderate to vigorous activity can help modify disease progression at this disease stage in ADAD.

The direction of the relationship between training load and GFAP levels was unexpected. Another study reported similar findings, in which it was shown that higher cardiorespiratory fitness was associated with higher NfL and GFAP levels in cognitively unimpaired older apolipoprotein E ε4 non-carriers.⁴⁵ GFAP is a marker of astrocyte reactivity, and reactive astrocytes can exacerbate or suppress neuropathology.⁴⁶ Astrocytes are activated or recruited to the site of injury or pathology to assist with Aβ clearance and degradation and help maintain homeostasis when the disease is developing.^47,48 That is, their function and response to protein misfolding and aggregation are considered neuroprotective or healthy at first before becoming pro-inflammatory.^46–48 One hypothesis is that, because mutation carriers were 15 years on average from MCI onset and at an age when they are accumulating amyloid plaques,⁴ higher levels of GFAP may represent an astrocytic response to the accumulation of amyloid that could be the immune system's attempt to maintain homeostasis. However, this is purely speculative, and more research is needed to understand this relationship, including studies with larger samples that can replicate the findings.

Our study had several shortcomings. First, TRIMP values suggest that most participants engaged minimal time in at least moderate PA, limiting our ability to examine differences between low versus high PA. Additionally, most TRIMP studies have focused on athletes and used devices other than FitBit to measure HR. The FitBit has inherent limitations, as the accuracy of step counts and HR measurements can vary based on factors like skin tone or wearing the device loosely, which could impact steps and TRIMP values.^49,50 We also did not know if some participants were taking medications or substances that could have affected their HR while they were using FitBit. Further, our sample size was relatively small, and participants were mostly in the very early stages of preclinical AD, which limited variability in biomarker levels and cognitive test scores. Future studies should include a PA diary and individuals who engage in greater PA, as well as longitudinal data, in a larger sample to understand better the relationships among PA, biomarkers of AD progression, and cognition.

However, our study also has several strengths. First, we collected objective measures that capture distinct aspects of PA (i.e., locomotion and duration/intensity) for 14 days in individuals with ADAD at risk for dementia. We also report data on objective cognitive tests and an array of biomarkers that are known to be critical for brain health and cognitive functioning in AD and effective at identifying individuals at risk for dementia. Further, we studied individuals in the preclinical stage of AD and who have a well-characterized disease trajectory. Of utmost importance, our participants do not typically show evidence of vascular disease or have cardiovascular risk factors, which often confounds findings in studies with older adults.

ACKNOWLEDGMENTS

We thank all participants for contributing their time to this study and for their continued commitment to this research. This work was supported by the US National Institute on Aging (NIA) (grant number K23 AG061276) and the MGH Executive Committee on Research Fund for Medical Discovery awarded to E.G-V. Y.T.Q received funding from the US National Institutes of Health (NIH) Office of the Director (grant number DP5 OD019833), US NIA (grant number R01 AG054671, RM1NS132996), the MGH Executive Committee on Research (MGH Research Scholar Award), and the Alzheimer's Association (grant number AARGD-591030), the Alzheimer's Association, and Massachusetts General Hospital ECOR. Y.T.Q serves as a consultant for Biogen. Y.T.Q. and F.L. received funding from the US NIA (grant number RF1AG077627). F.L. received funding from the NIH, Roche, the Banner Alzheimer's Foundation for the Alzheimer's Prevention Initiative (API) Colombia Registry, and the API ADAD Colombia Trial. A.R.-H. received funding from the US NIH NeuroID program (grant number R25 NS080687). Funding sources were not involved in the preparation of this article.

CONFLICT OF INTEREST STATEMENT

Quiroz and Lopera serve as consultants for Biogen. Guzmán-Vélez serves as a consultant for VarMed Management outside the submitted work. Arnold has consulted and/or served on advisory boards for Allyx Therapeutics, BioVie, Bob's Last Marathon, Daewoong Pharmaceuticals, Foster & Eldridge, LLP, Quince Therapeutics, Sage Therapeutics, and Vandria. He has received sponsored research grant support via his institution from the following commercial entities: AbbVie, Amylyx, Athira Pharma, Chromadex, Cyclerion Therapeutics, EIP Pharma, Ionis Pharmaceuticals, Janssen Pharmaceuticals, Inc., Novartis AG, Seer Biosciences, Inc., and vTv Therapeutics, Inc. He has received sponsored research grant support via his institution from the following non-commercial entities: Alzheimer's Association, Alzheimer's Drug Discovery Foundation, Challenger Foundation, Cure Alzheimer's Fund, John Sperling Foundation, the National Institutes of Health, and the Prion Alliance. All other co-authors have no conflicting interests to disclose. Author disclosures are available in the supporting information.

References

Alzheimer's Association. 2024 Alzheimer's Disease facts and figures. 2024;20(5).

Ismail R, Parbo P, Madsen LS, et al. The relationships between neuroinflammation, beta‐amyloid and tau deposition in Alzheimer's disease: a longitudinal PET study. J Neuroinflam. 2020;17(1):151.

Braak H, Del Tredici K. The preclinical phase of the pathological process underlying sporadic Alzheimer's disease. Brain. 2015;138(Pt 10):2814‐2833.

Quiroz YT, Sperling RA, Norton DJ, et al. Association between amyloid and tau accumulation in young adults with autosomal dominant Alzheimer disease. JAMA Neurol. 2018;75(5):548‐556.

Guzmán‐Vélez E, Diez I, Schoemaker D, et al. Amyloid‐β and tau pathologies relate to distinctive brain dysconnectomics in preclinical autosomal‐dominant Alzheimer's disease. Proc Natl Acad Sci USA. 2022;119(15): [eLocator: e2113641119].

Ising C, Venegas C, Zhang S, et al. NLRP3 inflammasome activation drives tau pathology. Nature. 2019;575(7784):669‐673.

Tejera D, Mercan D, Sanchez‐Caro JM, et al. Systemic inflammation impairs microglial Aβ clearance through NLRP3 inflammasome. EMBO J. 2019;38(17): [eLocator: e101064].

Caspersen CJ, Powell KE, Christenson GM. Physical activity, exercise, and physical fitness: definitions and distinctions for health‐related research. Public Health Rep. 1985;100(2):126‐131.

Livingston G, Huntley J, Sommerlad A, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. 2020;396(10248):413‐446.

Hörder H, Johansson L, Guo X, et al. Midlife cardiovascular fitness and dementia: a 44‐year longitudinal population study in women. Neurology. 2018;90(15):e1298‐e1305. doi: [DOI: https://dx.doi.org/10.1212/WNL.0000000000005290]

Erickson KI, Voss MW, Prakash RS, et al. Exercise training increases size of hippocampus and improves memory. Proc Nat Acad Sci USA. 2011;108(7):3017‐3022.

Voss MW, Prakash RS, Erickson KI, et al. Plasticity of brain networks in a randomized intervention trial of exercise training in older adults. Front Aging Neurosci. 2010;2:32.

Andel R, Crowe M, Pedersen NL, Fratiglioni L, Johansson B, Gatz M. Physical exercise at midlife and risk of dementia three decades later: a population‐based study of Swedish twins. J Gerontol A Biol Sci Med Sci. 2008;63(1):62‐66.

Vints WAJ, Seikinaite J, Gokce E, et al. Resistance exercise effects on hippocampus subfield volumes and biomarkers of neuroplasticity and neuroinflammation in older adults with low and high risk of mild cognitive impairment: a randomized controlled trial. Geroscience. 2024;46(4):3971‐3991.

Smith JC, Nielson KA, Woodard JL, et al. Physical activity reduces hippocampal atrophy in elders at genetic risk for Alzheimer's disease. Front Aging Neurosci. 2014;6:61.

Liang KY, Mintun MA, Fagan AM, et al. Exercise and Alzheimer's disease biomarkers in cognitively normal older adults. Ann Neurol. 2010;68(3):311‐318.

Lautenschlager NT, Cox KL, Flicker L, et al. Effect of physical activity on cognitive function in older adults at risk for Alzheimer disease: a randomized trial. JAMA. 2008;300(9):1027‐1037.

Buchman AS, Boyle PA, Yu L, Shah RC, Wilson RS, Bennett DA. Total daily physical activity and the risk of AD and cognitive decline in older adults. Neurology. 2012;78(17):1323‐1329.

Rabin JS, Klein H, Kirn DR, et al. Associations of Physical Activity and β‐Amyloid with longitudinal cognition and neurodegeneration in clinically normal older adults. JAMA Neurol. 2019;76(10):1203‐1210.

Okonkwo OC, Schultz SA, Oh JM, et al. Physical activity attenuates age‐related biomarker alterations in preclinical AD. Neurology. 2014;83(19):1753‐1760.

Muller S, Preische O, Sohrabi HR, et al. Relationship between physical activity, cognition, and Alzheimer pathology in autosomal dominant Alzheimer's disease. Alzheimers Dement. 2018;14(11):1427‐1437.

Prince SA, Adamo KB, Hamel ME, Hardt J, Connor Gorber S, Tremblay M. A comparison of direct versus self‐report measures for assessing physical activity in adults: a systematic review. Int J Behav Nutr Phys Act. 2008;5:56.

Aguirre‐Acevedo DC, Lopera F, Henao E, et al. Cognitive decline in a Colombian kindred with autosomal dominant Alzheimer Disease: a retrospective cohort study. JAMA Neurol. 2016;73(4):431‐438.

Master H, Annis J, Huang S, et al. Association of step counts over time with the risk of chronic disease in the All of Us Research Program. Nat Med. 2022;28(11):2301‐2308.

Banach M, Lewek J, Surma S, et al. The association between daily step count and all‐cause and cardiovascular mortality: a meta‐analysis. Eur J Prev Cardiol. 2023;30(18):1975‐1985.

Dempsey PC, Rowlands AV, Strain T, et al. Physical activity volume, intensity, and incident cardiovascular disease. Eur Heart J. 2022;43(46):4789‐4800.

Bassett DR Jr, Toth LP, LaMunion SR, Crouter SE. Step Counting: A review of measurement considerations and health‐related applications. Sports Med. 2017;47(7):1303‐1315.

Banister EW, Calvert TW. Planning for future performance: implications for long term training. Can J Appl Sport Sci. 1980;5(3):170‐176.

Garcia‐Ramos A, Feriche B, Calderon C, et al. Training load quantification in elite swimmers using a modified version of the training impulse method. Eur J Sport Sci. 2015;15(2):85‐93.

Mujika I. Quantification of training and competition loads in endurance sports: methods and applications. Int J Sports Physiol Perform. 2017;12(Suppl 2):S29‐S217.

Chatterjee P, Pedrini S, Doecke JD, et al. Plasma Abeta42/40 ratio, p‐tau181, GFAP, and NfL across the Alzheimer's disease continuum: a cross‐sectional and longitudinal study in the aibl cohort. Alzheimers Dement. 2023;19(4):1117‐1134.

Schindler SE, Bollinger JG, Ovod V, et al. High‐precision plasma beta‐amyloid 42/40 predicts current and future brain amyloidosis. Neurology. 2019;93(17):e1647‐e1659.

Shen XN, Huang YY, Chen SD, et al. Plasma phosphorylated‐tau181 as a predictive biomarker for Alzheimer's amyloid, tau and FDG PET status. Transl Psychiatry. 2021;11(1):585.

Guzmán‐Vélez E, Zetterberg H, Fox‐Fuller JT, et al. Associations between plasma neurofilament light, in vivo brain pathology, and cognition in non‐demented individuals with autosomal‐dominant Alzheimer's disease. Alzheimers Dement. 2021;17(5):813‐821.

Chatterjee P, Pedrini S, Stoops E, et al. Plasma glial fibrillary acidic protein is elevated in cognitively normal older adults at risk of Alzheimer's disease. Transl Psychiatry. 2021;11(1):27.

Quiroz YT, Zetterberg H, Reiman EM, et al. Plasma neurofilament light chain in the presenilin 1 E280A autosomal dominant Alzheimer's disease kindred: a cross‐sectional and longitudinal cohort study. Lancet Neurol. 2020;19(6):513‐521.

Vila‐Castelar C, Chen Y, Langella S, et al. Sex differences in blood biomarkers and cognitive performance in individuals with autosomal dominant Alzheimer's disease. Alzheimers Dement. 2023;19(9):4127‐4138.

Sclan SG, Reisberg B. Functional assessment staging (FAST) in Alzheimer's disease: reliability, validity, and ordinality. Int Psychogeriatr. 1992;4(Suppl 1):55‐69.

Lendon CL, Martinez A, Behrens IM, et al. E280A PS‐1 mutation causes Alzheimer's disease but age of onset is not modified by ApoE alleles. Hum Mutat. 1997;10(3):186‐195.

Hilden P, Schwartz JE, Pascual C, Diaz KM, Goldsmith J. How many days are needed? Measurement reliability of wearable device data to assess physical activity. PLoS One. 2023;18(2): [eLocator: e0282162].

Fuller D, Colwell E, Low J, et al. Reliability and validity of commercially available wearable devices for measuring steps, energy expenditure, and heart rate: systematic review. JMIR Mhealth Uhealth. 2020;8(9): [eLocator: e18694].

Garber CE, Blissmer B, Deschenes MR, et al. American College of Sports Medicine position stand. quantity and quality of exercise for developing and maintaining cardiorespiratory, musculoskeletal, and neuromotor fitness in apparently healthy adults: guidance for prescribing exercise. Med Sci Sports Exerc. 2011;43(7):1334‐1359.

Vidoni ED, Morris JK, Watts A, et al. Effect of aerobic exercise on amyloid accumulation in preclinical Alzheimer's: a 1‐year randomized controlled trial. PLoS One. 2021;16(1): [eLocator: e0244893].

Lamb SE, Sheehan B, Atherton N, et al. Dementia And Physical Activity (DAPA) trial of moderate to high intensity exercise training for people with dementia: randomised controlled trial. BMJ. 2018;361: [eLocator: k1675].

Sewell KR, Rainey‐Smith SR, Pedrini S, et al. The impact of exercise on blood‐based biomarkers of Alzheimer's disease in cognitively unimpaired older adults. Geroscience. 2024.

Jiwaji Z, Tiwari SS, Aviles‐Reyes RX, et al. Reactive astrocytes acquire neuroprotective as well as deleterious signatures in response to Tau and Ass pathology. Nat Commun. 2022;13(1):135.

Webers A, Heneka MT, Gleeson PA. The role of innate immune responses and neuroinflammation in amyloid accumulation and progression of Alzheimer's disease. Immunol Cell Biol. 2020;98(1):28‐41.

Kim J, Yoo ID, Lim J, Moon JS. Pathological phenotypes of astrocytes in Alzheimer's disease. Exp Mol Med. 2024;56(1):95‐99.

Dooley EE, Golaszewski NM, Bartholomew JB. Estimating accuracy at exercise intensities: a comparative study of self‐monitoring heart rate and physical activity wearable devices. JMIR Mhealth Uhealth. 2017;5(3): [eLocator: e34].

Lederer L, Breton A, Jeong H, Master H, Roghanizad AR, Dunn J. The importance of data quality control in using Fitbit device data from the research program. JMIR Mhealth Uhealth. 2023;11: [eLocator: e45103].