Association of Physical Activity With Neurofilament Light Chain Trajectories in Autosomal Dominant Frontotemporal Lobar Degeneration Variant Carriers

Key Points

Question

What is the association between physical activity levels and plasma neurofilament light chain (NfL) trajectories in autosomal dominant frontotemporal lobar degeneration (FTLD) variant carriers?

Findings

In this cohort study including 160 participants, FTLD variant carriers with higher baseline physical activity showed 14% and 30% slower NfL progression over 4 years compared with peers with average and low physical activity, respectively. The association between physical activity and NfL trajectories was strongest in C9orf72 and MAPT variant carriers and for activities with higher cardiorespiratory and cognitive demands.

Meaning

In this study, physical activity was associated with with axonal protection in those with FTLD.

This cohort study examines the association between physical activity and longitudinal neurofilament light chain (NfL) trajectories in individuals with autosomal dominant forms of frontotemporal lobar degeneration.

Abstract

Importance

Physical activity is associated with cognitive health, even in autosomal dominant forms of dementia. Higher physical activity is associated with slowed cognitive and functional declines over time in adults carrying autosomal dominant variants for frontotemporal lobar degeneration (FTLD), but whether axonal degeneration is a potential neuroprotective target of physical activity in individuals with FTLD is unknown.

Objective

To examine the association between physical activity and longitudinal neurofilament light chain (NfL) trajectories in individuals with autosomal dominant forms of FTLD.

Design, Setting, and Participants

This cohort study included individuals from the ALLFTD Consortium, which recruited patients from sites in the US and Canada. Symptomatic and asymptomatic adults with pathogenic variants in one of 3 common genes associated with FTLD (GRN, C9orf72, or MAPT) who reported baseline physical activity levels and completed annual blood draws were assessed annually for up to 4 years. Genotype, clinical measures, and blood draws were collected between December 2014 and June 2019; data were analyzed from August 2021 to January 2022. Associations between reported baseline physical activity and longitudinal plasma NfL changes were assessed using generalized linear mixed-effects models adjusting for baseline age, sex, education, functional severity, and motor symptoms.

Exposures

Baseline physical activity levels reported via the Physical Activity Scale for the Elderly. To estimate effect sizes, marginal means were calculated at 3 levels of physical activity: 1 SD above the mean represented high physical activity, 0 SD represented average physical activity, and 1 SD below the mean represented low physical activity.

Main Outcomes and Measures

Annual plasma NfL concentrations were measured with single-molecule array technology.

Results

Of 160 included FTLD variant carriers, 84 (52.5%) were female, and the mean (SD) age was 50.7 (14.7) years. A total of 51 (31.8%) were symptomatic, and 77 carried the C9orf72 variant; 39, GRN variant; and 44, MAPT variant. Higher baseline physical activity was associated with slower NfL trajectories over time. On average, NfL increased 45.8% (95% CI, 22.5 to 73.7) over 4 years in variant carriers. Variant carriers with high physical activity demonstrated 14.0% (95% CI, −22.7 to −4.3) slower NfL increases compared with those with average physical activity and 30% (95% CI, −52.2 to −8.8) slower NfL increases compared with those with low physical activity. Within genotype, C9orf72 and MAPT carriers with high physical activity evidenced 18% to 21% (95% CI, −43.4 to −7.2) attenuation in NfL, while the association between physical activity and NfL trajectory was not statistically significant in GRN carriers. Activities associated with higher cardiorespiratory and cognitive demands (sports, housework, and yardwork) were most strongly correlated with slower NfL trajectories (vs walking and strength training).

Conclusions and Relevance

In this study, higher reported physical activity was associated with slower progression of an axonal degeneration marker in individuals with autosomal dominant FTLD. Physical activity may serve as a primary prevention target in FTLD.

Introduction

Dementia evolves out of the interaction between inherited risk and life exposures. Although genes are determined, there are up to a dozen lifestyle factors implicated in brain health development across the life span, accounting for more than 40% of dementia risk.^1,2 Genetic cohorts, such as autosomal dominant forms of neurodegeneration, provide a unique opportunity to test the relative importance of nature vs nurture contributions to dementia risk. Recent work in autosomal dominant forms of Alzheimer disease (AD) and frontotemporal lobar degeneration (FTLD) demonstrate beneficial associations between active lifestyles and cognitive and functional outcomes,^3,4 similar to findings in sporadic disease.⁵ We previously showed that adults carrying autosomal dominant FTLD variants who reported higher baseline physical activity levels showed approximately 50% slower functional decline per year compared with peers with low activity.⁴ Further, physical activity attenuated the adverse association between frontotemporal atrophy and cognitive trajectories in FTLD variant carriers. FTLD is among the most common causes of dementia in adults younger than 65 years, and an estimated 30% of cases are familial,^6,7 underscoring the relevance and opportunity of understanding the impact of primary prevention factors in this disease. To extend more biologic understanding of these prior studies, we evaluated the association between physical activity and a molecular correlate sensitive to FTLD-related neurodegeneration—plasma neurofilament light chain (NfL).

NfL is an intermediate filament protein that contributes to axonal cytoskeletal structure and scaffolding for cytoplasmic organelle organization and facilitates microtubule dynamics for axonal transport.^8,9 NfL concentrations in plasma and cerebrospinal fluid are consistently reported as robust indicators of neuroaxonal damage across neurologic diseases and temporally predict future hypometabolic and cortical atrophy patterns in familial and sporadic AD.^10,11,12 In fact, plasma and cerebrospinal fluid NfL concentrations are particularly elevated in frontotemporal degeneration (FTD) syndromes, including amyotrophic lateral sclerosis, behavioral variant FTD, corticobasal syndrome, progressive supranuclear palsy, and semantic variant primary progressive aphasia compared with other neurodegenerative diseases (eg, AD, dementia with Lewy bodies).^13,14 Recent work describing the natural history of disease progression in FTLD variant carriers shows that NfL rises insidiously decades before symptom onset, with steep increases around the time of phenoconversion.^15,16,17 In an era without in vivo FTLD pathology biomarkers (ie, transactive response DNA-binding protein 43 [TDP-43] and FTLD-tau), NfL is among the most sensitive molecular indicators of FTLD-related disease processes. Thus, examining the association between a modifiable factor, physical activity, and NfL concentrations in individuals with FTLD will provide highly clinically relevant insights into intervenable targets for FTLD prevention.

We examined how baseline physical activity levels are associated with longitudinal NfL trajectories in plasma among autosomal dominant FTLD variant carriers from the ALLFTD Consortium. To determine the characteristics of this association, we probed the impact of FTLD variant genotype and types of physical activities most strongly predictive of NfL trajectories. Delineation of dynamic molecular markers associated with lifestyle factors will facilitate risk stratification and monitoring for more precise implementation of behavioral interventions and help inform pathophysiologic insights for FTLD.

Methods

Participants

Participants included individuals affected by genetic forms of FTLD enrolled in the Advancing Research and Treatment for Frontotemporal Lobar Degeneration (ARTFL) and Longitudinal Evaluation of Familial Frontotemporal Dementia (LEFFTDS) Longitudinal FTD study (ALLFTD)¹⁸ based in the US and Canada. Participants were included based on presence of one of the 3 most common FTLD genetic variants (MAPT, GRN, and C9orf72) and completion of baseline report of physical activities and blood draw within 90 days. Using these criteria, 160 individuals were included (Table 1). Most variant carriers (107 of 160 [66.9%]) were asymptomatic to mildly symptomatic at baseline (CDR Dementia Staging Instrument [CDR] plus National Alzheimer Coordinating Center [NACC] Behavior and Language Domains for FTLD [CDR+NACC FTLD] score less than 1). The ALLFTD study is an ongoing longitudinal study with approximately annual visits. All participants completed a baseline visit and averaged 2 annual visits (106 had more than 2 visits); participants with only 1 visit were slightly older (mean age, 54.6 vs 48.9; P = .04) and more functionally impaired on CDR+NACC FTLD (mean score, 0.85 vs 0.55; P = .03) at baseline compared with those with more than 2 visits but did not differ on sex (female sex, 31 of 54 [57.4%] vs 54 of 106 [50.9%]; P = .48) or education (mean years of education, 15.4 vs 15.3; P = .82). All genetic testing was completed in the same laboratory at the University of California, Los Angeles, using standardized methods previously described.^19,20 All participants provided written informed consent, and the study was approved by local institutional review boards. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Table 1. Clinical and Demographic Characteristics of Frontotemporal Lobar Degeneration Variant (FTLD) Carriers.

Characteristic	No. (%)
Total, No.	160
Visits
≥1	160 (100)
≥2	106 (66.3)
≥3	66 (41.3)
4	35 (21.95)
Age, mean (SD), y	50.7 (14.7)
Sex
Female	84 (52.5)
Male	76 (47.5)
Education, mean (SD), y	15.3 (2.5)
Genotype
C9orf72	77 (48.2)
GRN	39 (24.3)
MAPT	44 (27.5)
CDR+NACC FTLD SB, median (IQR; range)	0 (0-0.75; 0-3)
PASE, median (IQR)
Total score	107.8 (65.3-146.2)
Housework	50 (25-50)
Yardwork	0 (0-36)
Sports	4.8 (0-17.3)
Walking	15 (5-25.8)
Strength training	0 (0-7.5)
Neurofilament light chain, median (IQR), pg/mL
Mean	10.1 (5.3-24.4)
Coefficient of variance, %	3.8 (1.8-6.9)

Abbreviations: CDR+NACC FTLD SB, CDR Dementia Staging Instrument plus National Alzheimer Coordinating Center Behavior and Language Domains for Frontotemporal Lobar Degeneration sum of boxes; PASE, Physical Activity Scale for the Elderly.

Baseline Physical Activity

Participants completed a self-reported measure of physical activity (Physical Activity Scale for the Elderly [PASE]).²¹ The PASE is an 11-item measure of physical activity levels across life domains (eg, recreational, household) over the past 7 days. Participants rated weekly frequency and daily duration for the following recreational activities: (1) walking; (2) light, moderate, and strenuous sports; (3) housework; (4) yardwork; and (5) strength training. For each activity, a score was obtained by multiplying activity frequency by a task-specific weight according to the scoring manual. Activity scores were then summed to calculate a total PASE score representing overall physical activity level. Values ranged from 0 to more than 500, with higher values indicating greater levels of physical activity.²² The PASE has been widely validated for use in older adults across cultures and demonstrated adequate test-retest reliability (intraclass correlation coefficient, 0.49 to 0.95),^23,24,25 intraclass correlation (α= 0.65),²⁶ and construct validity.^21,24

Plasma Biomarker Quantification

Venous blood was collected in ethylenediaminetetraacetic acid–containing tubes. Blood was centrifuged at 1500 g and 4 °C for 15 minutes. Plasma samples were aliquoted in 1000-µL polypropylene tubes and stored at −80 °C, until further use. Samples (1 thaw only) were gradually brought to room temperature before analysis. NfL concentrations were quantified in duplicate using the ultrasensitive HDX analyzer by single-molecule array (Simoa) technology (Quanterix) by investigators blinded to clinical group allocation.¹⁷ Samples with coefficients of variance greater than 20% were excluded from analyses.

CDR+NACC FTLD

The CDR+NACC FTLD^19,27 was used as a marker of clinical severity and includes ratings across 6 functional domains captured in the traditional CDR in addition to 2 domains specific to the core clinical features of FTLD (language and behavior). Following a standardized algorithm,^19,27 the 8 domain scores were summed to create a global score (scale from 0 to 8), while each domain was scored on a scale from 0 to 3 and summed to create a more continuous measure of symptom severity (scale from 0 to 24), referred to as the sum of boxes (CDR+NACC FTLD SB).

Unified Parkinson Disease Rating Scale

The Unified Parkinson Disease Rating Scale (UPDRS) is a clinician-rated scale for parkinsonism.²⁸ The UPDRS motor examination consists of 27 items following a structured neurologic examination, with scores ranging from 0 (normal) to 4 (severely impaired) for a maximum total score of 108 points.

Statistical Analysis

Baseline Analyses

All analyses were conducted in JMP version 16 (JMP Statistical Discovery) or Stata version 15 (StataCorp). Statistical significance was set as 2-tailed α less than .05. We examined baseline associations between reported physical activity levels (PASE) and clinicodemographic features and plasma NfL concentrations in variant carriers using correlation and linear regression. Plasma NfL concentration was log transformed due to a significant positive skew. Regression models tested the association between physical activity and plasma NfL concentration, adjusting for age, sex, education, and clinical severity (CDR+NACC FTLD SB).

Longitudinal Models

Baseline PASE values were converted to a sample-based z score metric to enhance interpretability for longitudinal models. Given the significant skew of our dependent variable, NfL, and after examining Q-Q plots with log-transformed NfL, which evidenced improvement in skew and kurtosis, we opted to conduct generalized linear mixed-effects (GLMMs) models using the log function. We examined the association between baseline physical activity on longitudinal NfL trajectories by entering in the interaction between baseline physical activity and time in study (years from baseline PASE), adjusting for baseline age, sex, education, and CDR+NACC FTLD SB (time varying). In GLMMs, all time points with available data are incorporated to estimate parameters with person-specific intercepts specified. Intercepts adjust for person-specific baseline levels of NfL, which are strong predictors of clinical outcomes.^15,16,17 Physical activity levels may be affected by motor and cognitive/behavioral declines related to FTD disease severity (reverse causality). To probe whether associations were independent of clinical symptoms, we further adjusted for motor symptoms (UPDRS) and excluded variant carriers who were clinically symptomatic at baseline (CDR+NACC FTLD score greater than 0.5).

Post Hoc Models

We tested post hoc models to examine the characteristics (impact of genotype or activity type) of the association between baseline physical activity and NfL trajectories in variant carriers. Again, using log function GLMMs adjusting for demographic characteristics and CDR+NACC FTLD SB, we tested the 3-way interaction between baseline physical activity × time × genotype (C9orf72, GRN, and MAPT). Significant interaction models were stratified by genotype. Additionally, we examined the constituent components of the PASE scale: walking, sports (light, moderate, and strenuous combined), yardwork, housework, and strength training. Individual activity types were entered into GLMMs predicting NfL trajectories.

Effect sizes for log-transformed GLMMs were estimated using exponential b ratios (proportions) to enhance interpretability (reported as exp[b]). Values are reported as proportions and 95% CIs or SEs. We opted to report effect sizes estimated for a 4-year epoch, given that this may be a more clinically meaningful time frame to estimate progression (vs 1 year) that does not extrapolate past available data. To do so, years in study was divided by 4. To estimate effect sizes, marginal means were calculated at 3 levels of physical activity: 1 SD above the mean represented high physical activity (PASE score of 173), 0 SD represented average physical activity (PASE score of 111), and 1 SD below the mean represented low physical activity (PASE score of 49).

Results

Baseline Models

Of 160 included FTLD variant carriers, 84 (52.5%) were female, and the mean (SD) age was 50.7 (14.7) years. A total of 51 (31.8%) were symptomatic, and 77 carried the C9orf72 variant; 39, GRN; and 44, MAPT. In variant carriers, greater reported physical activity correlated with younger age (r = −0.20), male sex (t = −2.1), lower CDR+NACC FTLD SB (r = −0.29), and lower baseline plasma NfL concentration (r = −0.30). Reported physical activity levels did not significantly differ by genotype (F_3,157 = 1.73; P = .18; C9orf72: mean [SD], 115.0 [65.1]; GRN: mean [SD], 97.0 [60.2]; MAPT: mean [SD], 121.7 [61.6]). Adjusting for age, sex, education, and CDR+NACC FTLD SB, greater physical activity was associated with lower concentrations of plasma NfL (β = −0.13; b = −0.002; SE = 0.001; P = .03).

Longitudinal Models

We next evaluated the association between baseline physical activity and longitudinal plasma NfL trajectories. Greater baseline physical activity was associated with slower NfL trajectories in FTLD variant carriers (Table 2; Figure 1). Among those with average physical activity levels (PASE z score = 0), NfL was estimated to increase 45.8% (95% CI, 22.5-73.7) over 4 years. However, variant carriers reporting high levels of baseline physical activity evidenced a 14.0% (95% CI, −22.7 to −4.3) slower increase in NfL over 4 years compared with average trajectories. Compared with variant carriers with low physical activity, high activity carriers evidenced 30.3% (95% CI, −52.2 to −8.8) slower NfL progression over 4 years. These associations remained statistically significant further adjusting for motor symptoms (n = 154; −12.1%; SE, 0.05; P = .02). Restricting the analysis to variant carriers who were only asymptomatic or mildly symptomatic at baseline (baseline CDR+NACC FTLD of 0.5 or less; n = 107) did not substantially change the effect size of physical activity on NfL trajectories, but the association no longer reached statistical significance (−13.9%; SE, 0.06; P = .07).

Table 2. Mixed-Effects Models Examining the Association Between Baseline Physical Activity and Longitudinal Plasma Neurofilament Light Chain Trajectories in Frontotemporal Lobar Degeneration Variant (FTLD) Variant Carriers.

Measure	Exponential b (95% CI)	P value
Baseline age	1.04 (1.03-1.05)	<.001
Education	1.01 (0.96-1.05)	.74
Sex	1.09 (0.87-1.36)	.74
CDR+NACC FTLD SB	1.02 (1.01-1.02)	<.001
Baseline PASE (z score)	0.86 (0.77-0.96)	.005
Time (4 y)	1.46 (1.22-1.74)	<.001
Baseline PASE × time	0.86 (0.77-0.96)	.006

Abbreviations: CDR+NACC FTLD SB, CDR Dementia Staging Instrument plus National Alzheimer Coordinating Center Behavior and Language Domains for Frontotemporal Lobar Degeneration sum of boxes; FTLD, frontotemporal lobar degeneration; PASE, Physical Activity Scale for the Elderly.

Figure 1. Baseline Physical Activity Levels and Rate of Plasma Neurofilament Light Chain (NfL) Progression in Frontotemporal Lobar Degeneration Variant Carriers.

To estimate effect sizes, marginal means were calculated at 3 levels of physical activity: 1 SD above the mean represented high physical activity (Physical Activity Scale for the Elderly [PASE] score of 173), 0 SD represented average physical activity (PASE score of 111), and 1 SD below the mean represented low physical activity (PASE score of 49). The shaded areas indicate 95% CIs.

Post Hoc Models

Genotype

Within variant carriers, there was an interaction between physical activity and genotype on plasma NfL trajectories (eTable 1 in Supplement 1). Pairwise comparisons suggested that GRN variant carriers (n = 39) demonstrated the weakest association between physical activity and NfL trajectories compared with MAPT variant carriers (n = 44; exp[b] = 0.1.42; SE, 0.20; P = .01) or C9orf72 variant carriers (n = 77; exp[b] = 0.58; SE, 0.16; P < .001). MAPT and C9orf72 variant carriers did not statistically differ from one another (exp[b] = −0.85; SE, 0.09; P = .14). Stratifying by genotype, effect sizes estimated that high physical activity was significantly associated with a 21.0% (95% CI, −43.4 to −8.3) and 18.4% (95% CI, −28.1 to −7.2) slowing of NfL concentration increases in C9orf72 and MAPT variant carriers, respectively, over 4 years compared with average NfL trajectories (Figure 2). Physical activity did not associate with NfL trajectories in GRN variant carriers (7.9%; 95% CI, −21.0 to 47.6; P = .63).

Figure 2. Baseline Physical Activity Levels and Rate of Plasma Neurofilament Light Chain (NfL) Progression in Frontotemporal Lobar Degeneration Variant Carriers by Genotype.

To estimate effect sizes, marginal means were calculated at 3 levels of physical activity: 1 SD above the mean represented high physical activity (Physical Activity Scale for the Elderly [PASE] score of 173), 0 SD represented average physical activity (PASE score of 111), and 1 SD below the mean represented low physical activity (PASE score of 49). The shaded areas indicate 95% CIs.

Physical Activity Types

To determine the types of activities driving associations, we extracted the following activities from the PASE: walking, household work, yardwork, sports, and strength training activities. Higher participation in household work, yardwork, and sports were most strongly associated with NfL trajectories in FTLD variant carriers (Figure 3; eTable 2 in Supplement 1). Estimates suggested that variant carriers with high participation in these activities experienced a 12.9% (95% CI, −21.9 to −2.8; yardwork) to 19.6% (95% CI, −33.9 to −2.3; sport) slower NfL trajectory over 4 years compared with average trajectories. Walking (−10.7%; 95% CI, −21.4 to 3.8; P = .15) and strength training (2.8%; 95% CI, −25.5 to 23.4; P = .75) showed weaker and nonsignificant associations with NfL trajectories.

Figure 3. Baseline Physical Activity Types and Neurofilament Light Chain (NfL) Trajectories in Frontotemporal Lobar Degeneration Variant Carriers by Physical Activity.

To estimate effect sizes, marginal means were calculated at 3 levels of physical activity: 1 SD above the mean represented high physical activity (Physical Activity Scale for the Elderly [PASE] score of 173), 0 SD represented average physical activity (PASE score of 111), and 1 SD below the mean represented low physical activity (PASE score of 49). Error bars indicate 95% CIs.

Discussion

We found that greater baseline physical activity engagement was associated with overall lower levels and slower increases in plasma NfL concentrations over time in adults carrying autosomal dominant variants for FTLD. As an indicator of neuroaxonal degeneration, plasma NfL concentration is among the most robust molecular markers of FTD disease progression and is predictive of clinical phenoconversion,^15,17,29 underscoring the clinical relevance of our findings. Across variant carriers, high physical activity was associated with an estimated 14% slower NfL increase compared with average physical activity and 30% slower NfL increased compared with low physical activity over a 4-year period. The size of this association appeared to differ by genotype. The positive association between physical activity and NfL attenuation was strongest in C9orf72 variant carriers (21%) followed by MAPT carriers (18%) but was not significant in GRN carriers. When examining the physical activity characteristics underlying the association, most activities were implicated, but strongest associations were observed for activities with greater cardiorespiratory (sports) and cognitive (yardwork and housework) demands. Given the observational study design, we cannot rule out reverse causality (ie, greater impairment leading to higher NfL and less activity engagement); however, associations remained significant adjusting for functional severity (CDR+NACC FTLD SB) and motor features that could limit activity engagement (UPDRS). Similarly, effect sizes were similar when examining only asymptomatic or mildly symptomatic variant carriers (CDR+NACC FTLD of 0.5 or less), suggesting some independence from disease severity. Taken together, our findings suggest that physical activity participation is a modifiable factor associated with protection against FTLD-related axonal breakdown even among individuals carrying autosomal dominant genetic variants.

These results build on our prior study demonstrating that FTLD variant carriers with more active lifestyles evidence slower functional declines and better cognitive performances relative to their brain atrophy rates.⁴ At a molecular level, our findings suggest that these clinical associations may be explained, in part, by maintenance of axonal structure. Neurofilaments are critical regulators of microtubule dynamics and axonal caliber, which facilitate axoplasmic transport of mitochondria, lipids, synaptic vesicles, and other organelles from the cell body.^8,9 Although underlying mechanisms are not clear, our data suggest that activity may directly or indirectly support dynamic cytoskeletal maintenance, which can in turn facilitate the synaptic signaling benefits observed with exercise.^30,31 Of interest, we did not previously find a significant direct association between physical activity and gray matter volume trajectories in this cohort. Although longitudinal blood NfL levels track with cortical volumes,¹⁶ several studies suggest that axonal breakdown, and thus, NfL elevations may occur prior to fulminant neurodegeneration.^12,15 Most variant carriers in both our current and prior study were in asymptomatic to mildly symptomatic stages of disease (median CDR+NACC FTLD, 0). It is possible that NfL concentration is simply a more dynamic and sensitive indicator of cortical status, particularly in the earliest prodromal stages of disease. Nonetheless, these data contribute to the growing body of literature supporting an important role of modifiable behaviors, particularly physical activity, in the risk for both sporadic and autosomal dominant forms of dementia.

We additionally found the protective association between physical activity and NfL was disproportionately prominent in C9orf72 and MAPT variant carriers compared with GRN variant carriers. Syndromically, both C9orf72 and MAPT variants can cause primary motor syndromes (amyotrophic lateral sclerosis and corticobasal syndrome or progressive supranuclear palsy, respectively). At a systems level, physical activity may exert some of its protective effects by activating and training motor circuits vulnerable in these individuals. At a pathologic level, C9orf72 results in the misfolding and displacement of TDP-43 from the nucleus, whereas MAPT leads to tau misfolding. Our data therefore indicate that physical activity may be beneficial regardless of molecular pathophysiology, suggesting that specific protein aggregation may not be the primary pathway through which physical activity supports brain health. This is consistent with some previous studies in AD, which do not show a consistent or strong direct association between physical activity and amyloid or tau.^32,33,34 Rather, physical activity more consistently appears to moderate the neurotoxic association between AD pathology and clinical status.^32,33 For instance, recent work from our team identified significant moderation, but not direct effects, of physical activity on the association between brain tissue levels of AD or TDP-43 pathology and cognitive trajectories.³³ Disease-independent mechanisms through which physical activity is associated with brain health increasingly support immune/glial, vascular and synaptic pathways.^35,36,37 For instance, in humans, greater physical activity is associated with higher synaptic protein levels in brain tissue at autopsy, an association that appears to be related, at least in part, to reduced microglial activation.^38,39 Similarly, recent animal models show that systemic increases in complement and coagulation inhibitors (ie, Gpld1, clusterin) following exercise mediate its neurogenetic and behavioral benefits in aged mice.^40,41 Similar immune and glial functions are increasingly implicated in FTLD development.^42,43 In contrast, GRN variant carriers did not evidence a significant association between physical activity and NfL, which was unexpected. Notably, GRN carriers represented the smallest group (n = 39), and we may not have been powered to reliably estimate effects. Alternatively, GRN is associated with particularly aggressive elevations in NfL²⁹; it is also possible that the protective benefits of physical activity are not sufficient to overcome the neurotoxic virulence of GRN variants. Future analyses in larger cohorts, including sporadic FTD, are needed to carefully parse out the person-specific and disease-specific factors that predict who stands to benefit most from physical activity.

Post hoc analyses suggested that specifically sports, yardwork, and housework activities, characterized by greater cardiorespiratory and cognitive demands, may most strongly predict NfL trajectories in variant carriers. There is not yet evidence supporting the specific activity patterns that are most critical for brain health. A recent meta-analysis of exercise trials suggest that both high-impact and low-impact activities promote dose-dependent cognitive improvements in older adults, consistent with our findings.⁵ In particular, cardiorespiratory activities aimed at increasing maximal oxygen consumption have shown some of the most consistent beneficial effects for cognitive outcomes and future dementia risk.^44,45,46 We found variant carriers reporting higher levels of sporting activities (eg, tennis, skiing) showed an almost 20% slower increase in NfL over the next 4 years compared with peers with average levels of sporting activities. Nonetheless, it is notable that some of the earliest evidence linking physical activity with cognitive aging demonstrated beneficial associations even with low-impact activities, such as walking.⁴⁷ Indeed, yardwork and housework would not be expected to affect respiratory fitness, yet variant carriers engaging in these activities evidenced 13% to 18% attenuation in NfL concentrations over the next 4 years. Both yardwork and housework (as well as sports) may involve relatively greater cognitive complexity in conjunction with movement, which may promote brain health particularly compared with walking and strength training, which evidenced the smallest associations with NfL. Notably, activity types were self-reported and may have reduced reliability compared with the total score. Objective monitoring of physical activity via digital health tools (eg, actigraphy) are increasingly feasible and necessary to capture nuances of activity patterns, including duration, frequency, and intensities, most relevant for brain health.

Limitations

Our study has several important limitations. Our design was observational, and we cannot determine directionality. Although animal models and, increasingly, human randomized clinical trials demonstrate causal effects between exercise and neurogenic and cognitive improvements,^40,48,49,50 it is likely that these relationships are highly bidirectional. Additionally, physical activity levels were self-reported and only available at baseline. Self-reported physical activity only shows modest correlations (r of approximately 0.30) with objective actigraphy-based indicators and carries inherent limitations (eg, recall and social bias). Similarly, although the PASE total score is a well-supported metric of physical activity, the individual domain scores that we leveraged in our post hoc analyses have not been widely used, and their psychometric properties are not as well understood. In addition, only baseline physical activity levels were captured in the ALLFTD study; a deeper understanding of the temporally linked dynamics between changes in physical activity and changes in neurophysiologic and behavioral outcomes is needed to precisely determine when and how activity intervention may be most beneficial. Digital health monitoring tools are optimal for parsing these associations. Our sample sizes were relatively small, particularly within genotypes, and the follow-up period only spanned 4 years in a largely mildly affected cohort. It is possible that we were underpowered to detect some effects (eg, GRN carriers), and precision of our estimates (95% CIs) were relatively wide and need to be replicated. Notably, concentrations of blood-based biomarkers have been associated with body composition and, more recently, hepatorenal functioning.^51,52 Specifically, plasma NfL concentrations are increased in individuals with poor systemic clearance markers.^51,52 Body mass index and hepatorenal functioning were not systemically assessed in the ALLFTD study, and their potential confounding effect is therefore unknown. Peripheral health is an increasingly important factor to consider in future dementia studies of blood-based biomarkers.

Conclusions

In sum, baseline physical activity level was associated with up to 30% attenuation in plasma NfL accumulation in autosomal dominant FTLD variant carriers over 4 years. Although it is a relatively rare neurodegenerative disease, FTD is among the most common form of early-age dementia. To our knowledge, there have not been any reported lifestyle intervention trials in FTD to date. Given the complexity of FTD pathogenesis, combination therapies, including both pharmacological and behavioral interventions, may be needed. Phenotyping the neuroprotective behaviors in FTD is therefore an essential area of study. Identification of the molecular correlates of protective lifestyle behaviors will help elucidate common mechanisms supporting brain health and support a more precise roadmap (eg, risk stratification, complementary targets, response monitoring) for therapy implementation.

Supplement 1.

eTable 1. Generalized Linear Mixed-Effects Models Examining the Relationship Between Baseline Physical Activity and Longitudinal Plasma NfL Trajectories by FTLD Variant Genotypes

eTable 2. Generalized Linear Mixed-Effects Models Examining the Relationship Between Baseline Physical Activity and Longitudinal Plasma NfL Trajectories by Activity Type in FTLD Variant Carriers

Supplement 2.

Nonauthor Collaborators. ALLFTD Consortium