Moderating role of physical activity on hippocampal iron deposition and memory outcomes in typically aging older adults

Physical activity (PA) is linked to better cognitive and brain health, though its mechanisms are unknown. While brain iron is essential for normal function, levels increase with age and, when excessive, can cause detrimental neural effects. We examined how objectively measured PA relates to cerebral iron deposition and memory functioning in normal older adults. Sixty-eight cognitively unimpaired older adults from the UCSF Memory and Aging Center completed neuropsychological testing and brain magnetic resonance imaging, followed by 30-day Fitbit monitoring. Magnetic resonance imaging quantitative susceptibility mapping (QSM) quantified iron deposition. PA was operationalized as average daily steps. Linear regression models examined memory as a function of hippocampal QSM, PA, and their interaction. Higher bilateral hippocampal iron deposition correlated with worse memory but was not strongly related to PA. Covarying for demographics, PA moderated the relationship between bilateral hippocampal iron deposition and memory such that the negative effect of hippocampal QSM on memory performances was no longer significant above 9120 daily steps. PA may mitigate adverse iron-related pathways for memory health.


Introduction
Cognitive aging trajectories are heterogeneous and complex, ranging from stability and even increasing skills to significant decline and dementia.Despite its widespread impact as a leading cause of chronic disability and dependency in older adults, dementia does not represent typical changes in the biological aging process and is not an inevitable consequence of aging (Qiu and Fratiglioni, 2018).Thus, the identification of factors that impact cognitive aging outcomes, particularly those that may be modifiable, is an important area of research.The alterable nature of lifestyle behaviors makes them a promising target for primary prevention against cognitive impairment, particularly in the absence of available efficacious disease-modifying treatments for all-cause dementia.A recent cross-sectional study of risk factor prevalence in US adults cited that 41.0% of the nationwide dementia population has a diagnosis associated with modifiable lifestyle factors (Lee et al., 2022).Of the 12 lifestyle indices assessed, one of the top risk factors was physical inactivity.Physical activity (PA) is consistently associated with greater cognitive performance and brain aging outcomes (Barnes et al., 2003), as well as reduced incidence of dementias such as Alzheimer's disease (AD) (Buchman et al., 2012).As such, PA is widely supported as a lifestyle modification for older adults (Livingston et al., 2020).However, the precise biological mechanisms linking PA and exercise to brain health are still under investigation.Identification and thorough understanding of these biological mechanisms are essential to develop precise risk-stratification approaches for PA recommendations and identifying novel therapy targets (Möller et al., 2019).
A key measure of interest that may contribute to our understanding of the neurobiological mechanisms of PA is brain-based iron accumulation.As the most abundant paramagnetic agent in the human brain, iron plays a critical role in normal brain function.Iron is essential for brain homeostasis, including oxidative metabolism, formation and maintenance of neural networks, and myelin synthesis (Acosta-Cabronero et al., 2013).Importantly, increased brain iron deposition with age has been evidenced in both in vivo cross-sectional (Daugherty and Raz, 2013;Haacke et al., 2005) and longitudinal studies (Daugherty et al., 2015;Daugherty and Raz, 2016).While brain iron accumulation occurs during typical aging, dysregulation of iron can lead to highly elevated levels with detrimental effects.Excessive iron accumulation is thought to encourage spontaneous, neurotoxic release of free iron, which can then catalyze the formation of highly reactive radical species.This process is associated with exacerbated oxidative stress and increased cell predisposal to neuronal death (Acosta-Cabronero et al., 2016), as well as the propagation of myelin breakdown and neurodegeneration (Khattar et al., 2021).
Through a variety of magnetic resonance imaging (MRI)-based methods, advancing technology has enabled the ability to quantify brain iron deposition in vivo, especially as it relates to cognition.Cerebral iron deposition accumulation with age has been shown to be associated with poorer cognitive outcomes.MRI-based estimates of brain iron concentrations correlate with age-related differences in cognition cross sectionally (Hosking et al., 2018), and such findings are also reflected in longitudinal studies of cognitive decline in aging (Daugherty et al., 2015).Indeed, elevated cerebral iron load, particularly in the hippocampus, has been associated with poorer overall cognitive performances (Chen et al., 2021) and predicts accelerated decline on tests of memory and executive functions even in cognitively normal older adults (Zachariou et al., 2020).Associations are also demonstrated between hippocampal iron deposition and declarative memory.In healthy older adults, lower declarative memory scores have been linked to both higher hippocampal iron concentration and smaller hippocampal volume (Rodrigue et al., 2013), while longitudinal spatial memory improvement was predicted by lower baseline hippocampal iron levels and larger parahippocampal volumes (Daugherty and Raz, 2017).In clinical comparative studies, hippocampal iron deposition distinguished AD patients from healthy controls as measured through quantitative phase imaging (Ding et al., 2009) and a combination of ferritin iron accumulation and associated hippocampal tissue damage (Raven et al., 2013).Iron has been recognized as a significant source of oxidative stress that contributes to AD progression, and disruption to brain iron metabolism is postulated to influence AD pathogenesis (Honda et al., 2004).Based on supporting hippocampal iron accumulation and memory relationships and associations between altered brain iron metabolism and neurodegenerative disease (Schenck and Zimmerman, 2004), we pursued a study of brain iron deposition as a biomarker for neural and cognitive decline using quantitative susceptibility mapping (QSM).Major reservoirs of iron in the human brain including ferratin and neuromelanin pigments create local field distortions in the presence of a magnetic field that can be detected at submillimeter resolutions (Möller et al., 2019).This novel, MRI-based technique sensitively measures in vivo iron deposition via relaxation and magnetic susceptibility of brain tissue.
As an effect of its metabolic activity, density of oxidizable substrates, and generally low antioxidant defense, the brain is highly susceptible to oxidative stress (García-Mesa et al., 2016).PA, on the other hand, is linked to reduced oxidative stress, greater white matter integrity, and maintained myelin content (El Assar et al., 2022).Regular PA has been shown to increase cell and tissue endurance to oxidative stress, vascularization, and neurotrophin synthesis, processes that are closely related to neurogenesis, memory improvement, and brain plasticity (Radak et al., 2010).By promoting increased activity of enzymatic antioxidants, PA leads to heightened resistance to oxidative stress-related diseases, including cardiovascular and neurodegenerative disorders such as AD and Parkinson's disease (Kim et al., 2015).Despite the overlap of potential neurobiological targets, it is currently unknown how PA may associate with age-related iron deposition and related cognition changes.
To explore this relationship, our study evaluated 68 cognitively normal older adults who completed 30 days of Fitbit monitoring, neuropsychological assessment, and brain MRI including QSM and diffusion tensor imaging (DTI).Our primary study aim was to characterize associations among hippocampal iron deposition, memory outcomes, and PA in a cohort of cognitive unimpaired older adults.We hypothesized that (1) greater hippocampal iron deposition would correlate with worse memory outcomes and (2) increased levels of PA would directly relate to lower iron deposition and/or that increased levels of PA would attenuate the negative relationship between agerelated iron deposition and memory performance.

Participants
Participants included 68 community-dwelling, cognitively unimpaired older adults ages 55 and older that were enrolled in the UCSF Memory and Aging Center's Brain Aging Network for Cognitive Health (BRANCH).All participants received a clinical dementia rating of 0. To be included in the current study, participants completed neuropsychological testing and brain MRI scans, followed by Fitbit PA monitoring for 30 days.Participants were medically screened for a history of the following conditions: drinking two or more alcoholic beverages a night, substance abuse, epilepsy, brain tumors, multiple sclerosis, Parkinson's disease, post-traumatic stress disorder, schizophrenia, sleep apnea, major depression, anxiety disorder, head injury with loss of consciousness, surgery requiring anesthesia in the past 6 months, stroke, cancer (and treatment), eye surgery, severe vision deficits (e.g., cataracts, macular degeneration), colorblindness, high blood pressure, high cholesterol, and diabetes.The aforementioned cardiovascular health risk factors were noted in the participant's health history but were not exclusionary to participation.Brain Aging Network for Cognitive Health exclusion criteria included history or current evidence of the following conditions: large vessel stroke, diagnosis of DSM-5 major psychiatric disorders, multiple sclerosis, symptomatic neurodegenerative disease (e.g., Parkinson's disease), epilepsy, significant memory concerns or related diagnoses, active substance abuse, hepatitis C, HIV, syphilis, blindness, or deafness.This study was approved by the UCSF Institutional Review Board and all participants provided written, informed consent.

Participant characteristics
In the study sample of 68 cognitively unimpaired older adults, 56% were female and 85% identified as non-Hispanic White.On average, participants were 78.5 years old and reported 18.2 years of education.Participants averaged 7319 steps per day.Additional participant demographic characteristics are displayed in Table 1.

Fitbit metrics
All participants wore a Fitbit Flex 2 for 30 continuous days following research visits.Fitbit monitoring was conducted observationally during waking hours.All Fitbit accounts were linked to Fitabase, a platform specifically tailored to wearable research data management.All de-identified participant Fitbit data were then exported from Fitabase and quality-checked, which included removing individual days with fewer than 100 steps from analyses to control for nonadherence and only including participant monitoring data if there were ≥14 days of available data (Paolillo et al., 2022;VandeBunte et al., 2022).Day-level step count data were aggregated into the mean total daily steps for each participant.

Quantitative susceptibility mapping
All participants completed a brain MRI using a Siemens Prisma 3T scanner located at the UCSF Neuroscience Imaging Center.The calculation of QSM necessitated the acquisition of gradient-echo sequences using fast low-angle shot with a series of 8 echo times T2 * [4,9,14,19,24,29,34,39] ms and a flip angle of 15°.The acquisition matrix used was 0.9 × 0.9 × 2.0 mm 3 .All the images were registered in a QSM group template using Advanced Normalization Tools group template registration (Avants et al., 2009).Every single subject transformation was checked in the group template.The ROIs were extracted from an atlas registered in the group template space.
To compute tissue magnetic susceptibility maps based on gradient echoes, we used the STI suite (Li et al., 2014).In summary, the application unwraps the measured phase images and removes contributions caused by background susceptibilities using a Laplacianbased method.A customized group template was generated from subject susceptibility maps by linear and nonlinear registration template generation using the large deformation diffeomorphic metric mapping framework (Avants et al., 2008).Native subjects' susceptibility maps were geometrically normalized to the group template and smoothed in the group template.The applied smoothing used a Gaussian kernel with 8 mm full width half maximum.Every step of the transformation was carefully inspected from the native space to the group template.Susceptibility values of regions of interest were averaged and extracted from the Desikan atlas.
Our primary QSM outcome of interest was bilateral hippocampal iron deposition, calculated as an average of lateral hippocampal iron deposition.Although there are a relatively limited number of studies specific to hippocampal iron deposition, increased understanding of brain iron metabolism, relationship of iron to neurodegeneration, and availability of novel iron-dependent MRI methods support using brain iron accumulation as a potential biomarker for cognitive decline (Schenck and Zimmerman, 2004).We chose the hippocampus as a target due to previously reported findings that higher levels of hippocampal iron deposition correlate with decreased memory performance (Spence et al., 2020) and randomized controlled exercise trials implicating the hippocampus in older adults (Chapman et al., 2013;Erickson et al., 2011).
To determine the specificity of hypothesized relationships to hippocampal QSM, we also calculated a secondary QSM composite reflecting subcortical iron deposition to serve as a control region.Subcortical iron deposition was computed by averaging lateral iron deposition regional values across the caudate, thalamus, putamen, and globus pallidus.

Structural neuroimaging
Given that iron deposition is linked to white matter dysregulation, we aimed to determine the specificity of QSM models, independent of white matter integrity.Therefore, DTI fractional anisotropy (FA) was used to adjust for white matter integrity in QSM analyses.
Diffusion-weighted images were acquired via a single-shot spinecho planar imaging sequence with the following parameters: 69 axial slices with in-plane resolution of 2.0 mm and slice thickness of 2.0 mm (isotropic voxel); TR/TE 2420/72.20 ms; flip angle = 85°, 2 volumes B = 0 s/mm 2 with opposite phase encoding (AP/PA), 10 volumes, and 3 multishells with 96 noncollinear diffusion sensitization directions at b = 2500 s/mm 2 , 48 directions at B = 1000 s/ mm 2 , and 30 directions at B = 500 s/mm 2 , with an integrated parallel acquisition techniques acceleration factor of 2 and multiband acceleration factor of 3.
We reconstructed the diffusion tensor images following principles from Basser and Pierpaoli (Basser and Pierpaoli, 1996).Diffusion imaging processing began with denoising (Veraart et al., 2016).Then, images were realigned to the primary volume of the sequence, using the FSL MCFLIRT algorithm (Jenkinson et al., 2002).Data reflecting absolute displacement parameters beyond 1 mm were screened out and removed if necessary.Background voxels not considered as brain tissue were then masked out of the DWI volumes by applying a median Otsu function (Otsu, 1979).This function utilized the B0 acquisitions to provide a mask using Otsu thresholding with a 4 mm radius and 4 iterations to minimize intraclass variance (Garyfallidis et al., 2014).We used the realigned diffusion images, the mask, and the b-vectors and b-values in the eddy current-induced distortions correction process (Andersson and Sotiropoulos, 2016).Angular parameters, output of the previous step, were used to correct the bvector directions.Diffusion tensors were then fitted using Dipy (Garyfallidis et al., 2014) with a nonlinear least-squares approach.To estimate diffusion, FA measures derived from the fitted tensor were reconstructed in the native space for quality control.Our primary DTI FA measures of interest were the uncinate fasciculus and fornix tracts, calculated as an average of the left and right tracts.
To account for potential relationships between brain volume and iron deposition, we also calculated hippocampal brain volumes for use in our post-hoc QSM models.All T1-weighted images were visually quality-checked and excluded for excessive motion or artifacts.Tissue segmentation was performed using SPM12's unified segmentation procedure (Penny et al., 2007).To create a studyspecific template for warping individual participant T1-weighted images, we employed diffeomorphic anatomical registration using exponentiated Lie algebra (DARTEL) (Ashburner, 2007).All images were normalized within the study-specific template space using nonlinear and rigid-body registration.Smoothing was performed using an 8 mm full-width half-maximum Gaussian kernel.To facilitate registration with a brain parcellation atlas, linear and nonlinear transformations between diffeomorphic anatomical registration using exponentiated Lie algebra's space and International Consortium of Brain Mapping space were applied.Volume quantification required transforming a standard parcellation atlas into International Consortium of Brain Mapping space and summing all gray matter within parcellated regions of interest (Desikan et al., 2006).Total intracranial volume (TIV) was calculated as the sum of gray matter, white matter, and cerebrospinal fluid.We then examined TIV-adjusted hippocampal volume as a ratio of gray matter hippocampal volume to TIV.

Memory performance
Participants completed a brief neuropsychological assessment battery.Given the previously reported utility of hippocampal QSM to detect changes in normal adults and reported relationships between PA and memory performances (Chieffi et al., 2017;Spence et al., 2020), we opted to focus on episodic memory as our primary cognitive outcome.Episodic memory was assessed via the number of words recalled on the long-delay free recall trial from the California Verbal Learning Test, Second Edition (CVLT-II) (Beck et al., 2012).Sample-based z-scores were calculated and used as the primary measure of memory.

Statistical analyses
First, Pearson correlations examined univariate associations of both bilateral and lateral hippocampal iron deposition with PA (daily steps) and memory (CVLT-II long-delay free recall z-score).Univariate bilateral hippocampal iron deposition associations with age, sex, education, and DTI (uncinate fasciculus and fornix FA tracts) were also tested.Next, linear regression models examined memory as a function of hippocampal iron deposition, PA, and their interaction, covarying for age, sex, and education.We performed standard data diagnostics and found our model residuals to be normally distributed, with no suggested evidence of nonlinear patterns in the data.Posthoc models were conducted to test both laterality and specificity of memory to hippocampal iron deposition.Left hippocampal, right hippocampal, and subcortical iron deposition were modeled, respectively.Secondary models covaried for hippocampal gray matter volume and FA levels of the uncinate fasciculus and fornix in order to account for possible effects of hippocampal gray and white matter integrity on hippocampal iron deposition.To follow up on interaction findings, we employed the Johnson-Neyman method (Johnson and Neyman,1936) to identify the specific level of PA at which the relationship between hippocampal iron deposition and CVLT-II longdelay free recall z-scores was no longer statistically significant (False Discovery Rate adjusted).
Adjusting for age, sex, and education, average daily steps significantly moderated the relationship between bilateral hippocampal iron deposition and memory performance (β = 0.37, p = 0.013, 95% [confidence interval] CI = [0.08,0.65]; Table 2), such that, at higher levels of PA, the negative relationship between hippocampal iron deposition and memory was significantly attenuated (Fig. 2A).To better characterize this attenuation, follow-up Johnson-Neyman analysis indicated that the negative relationship between bilateral hippocampal iron deposition and memory was reduced to a nonsignificant effect once a threshold of approximately 9000 average total daily steps (9120) was surpassed (Fig. 2B).
To understand the specificity of estimates to hippocampal iron deposition, an additional model examined subcortical iron deposition.Subcortical iron deposition did not correlate with memory performance (r = −0.10,p = 0.46) nor was this relationship moderated by average daily steps (β = 0.06, p = 0.63, 95% CI = [−0.20,0.33]).To additionally account for the potential influence of regional myelin levels and white matter integrity on hippocampal iron deposition, we further adjusted our bilateral model for DTI-measured integrity of the uncinate fasciculus and fornix white matter tracts.The significant interaction between average daily steps and bilateral hippocampal iron deposition on memory evidenced a similar effect size and persisted after covarying for DTI (β = 0.37, p = 0.014, 95% CI = [0.08,0.65]; Table 2).

Discussion
We found that hippocampal iron deposition was negatively associated with memory performances and that PA significantly attenuated this adverse relationship in clinically normal older adults.Multiple R 2 = 0.3028, Adjusted R 2 = 0.1842 Key: DTI, diffusion tensor imaging; FA, fractional anisotropy.
Specifically, our analysis revealed that the adverse relationship between hippocampal iron deposition and memory performances only showed statistical significance below an average of 9120 average daily steps.This threshold falls within 7000-10,000 daily steps range that is consistently cited for its beneficial health outcomes, including cognitive aging (Del Pozo Cruz et al., 2022).Our sensitivity analyses also suggest that results were not driven by lateralized hippocampal iron deposition, were specific to iron deposition in the hippocampus, and were statistically independent of white matter integrity.Interestingly, we did not observe a strong direct relationship between PA levels and QSM, suggesting that activity may not be directly impacting the production or clearance/metabolism of cerebral iron deposition.Nonetheless, ours are the first data to our knowledge characterizing the in vivo relationships among cerebral iron deposition, objectively quantified PA, and cognition.We demonstrate the importance of targeting PA as a modifiable lifestyle factor that may help attenuate iron-related mechanisms of cognitive decline and serve as a primary prevention and/or behavioral intervention tool for individuals at risk for memory decline.
Highly elevated brain iron deposition has been shown to promote the formation of reactive oxygen species (ROS) and act as a catalyst for oxidative stress and neuronal apoptosis (Auten and Davis, 2009).Based on our findings, the primary role of PA does not appear to be directly related to iron deposition levels.PA may instead function by attenuating the detrimental cognitive outcomes associated with iron deposition.Although our data do not directly measure the molecular mechanisms underlying this attenuation effect, literature suggests that PA may stimulate the production of antioxidant promoters and myokines that protect against and/or respond to deposition-related oxidative stress (El Assar et al., 2022).For example, among a host of immune signals, exercise has been reported to upregulate nuclear factor erythroid 2-like 2, a major mediator of inflammation resolution (Sandberg et al., 2014).Following oxidative stress events, nuclear factor erythroid 2-like 2 induces the expression of antioxidants and cytoprotective genes to protect against ROS-induced damage and provoke anti-inflammatory responses against stressors (Vomund et al., 2017).In addition, irisin/FNDC5, an anti-inflammatory myokine expressed in skeletal muscle and brain tissue, is induced by PA.Irisin/FNDC5 also functions to decrease ROS production (Mancinelli et al., 2021) and has been shown to protect against metabolic stressors, including oxidative stress (Mazur-Bialy et al., 2018).Based on its ability to down-regulate ROS production in the brain and increase the level/activity of antioxidant enzymes in different brain regions, this may be a plausible mechanism by which PA can significantly attenuate oxidative stress in the brain and iron deposition-related homeostasis challenges (Radak et al., 2010;Simioni et al., 2018).PA has been shown to improve antioxidant defenses in older adult individuals to levels comparable with young, sedentary subjects (Bouzid et al., 2018), emphasizing the potential of PA to mitigate age-associated impairments tied to iron accumulation, including cognition and memory.Through these means, PA may potentially offset iron deposition-related cognitive disruption by moderating the brain oxidative stress pathway.
Unchecked iron deposition is also hypothesized to detrimentally impact myelination (Todorich et al., 2009).While oligodendrocyte myelin production and maintenance do require consistent iron supply (Bartzokis et al., 2007), continual, substantial brain iron accumulation is associated with oxidative stress that promotes neurodegeneration and myelin breakdown as observed in AD (Bartzokis, 2011).Oxidative injury and cyclic myelin degradation are further amplified when iron is released from oligodendrocytes during active demyelination (Haider et al., 2014).Although white matter integrity is known to decrease in both typical aging and dementia processes (Madden et al., 2012), recent randomized controlled trials demonstrated exercise-related effects on increased white matter plasticity (Mendez Colmenares et al., 2021) and integrity of white matter tracts (Bashir et al., 2021).In a recent cross-sectional observational study of older adults with cerebral small vessel disease and mild cognitive impairment, greater PA was linked to higher myelin content in whole-brain white matter (Boa Sorte Silva et al., 2023).Preclinical animal studies of myelin have found that increased exercise can be beneficial for axonal myelination, including increasing myelin sheath maturation, thickness, and potentially regeneration (Feter et al., 2018), which we hypothesize may be true in humans.In the presence of iron deposition, targeting oxidative stress or myelin degeneration may therefore be potential mechanisms through which PA may moderate detrimental effects on memory performance.We attempted to account for myelin health by covarying our primary model for DTI FA.Given that we found the moderating effect of PA persisted even adjusting for DTI FA, our data suggest that PA relationships may be at least in part independent of gross white matter structure.Future studies capturing molecular markers of myelin health may help to parse out these hypothesized mechanisms.
This study also highlights important clinical implications for both MRI QSM and the application of PA for cognitive health.Given the observed specific relationship between hippocampal QSM and memory performance in unimpaired older adults, iron deposition measures through QSM may potentially serve as a preclinical indicator of neurodegeneration risk, consistent with emerging studies on this relatively novel MRI tool (Wang et al., 2017).Our findings in clinically normal older adults also underscore the potential utility of PA as a primary prevention tool against age-related cognitive changes.PA may serve to mitigate the effects of elevated iron deposition on related cognitive changes, even prior to any form of clinical symptom onset.Taken together, these findings may inform a precision medicine approach to PA recommendations based on brain iron deposition levels.
Our study was not without limitations.Our relatively small sample size (n = 68) likely limited statistical power.This, as well as a restricted age range (56-93), may have impacted our ability to identify an expected relationship between hippocampal iron deposition and age.Considering our small sample size, we also wanted to keep the total number of comparisons in our models to a minimum and did not have robust data available for other measures of PA.This is a relevant area of study for future investigations.Given the observational nature of our study design, it is important to consider the role of reverse causality (i.e., accumulating iron deposition may result in lower PA engagement).It is likely that the relationship between PA and brain iron deposition levels is bidirectional with at least some contribution from reverse causality.Based on our cross-sectional design, it is not clear if iron deposition levels reflect longstanding brain differences that may impact subsequent PA engagement or vice versa.Future longitudinal studies will be essential for elucidating the temporal dynamics of how PA, iron deposition, and memory relate to one another over time.We also acknowledge that quantification of cerebral iron deposition and meaningful interpretation of QSM values are emerging fields of interest.Standardizing MRI-based QSM across scanners and study protocols is a barrier that has been reported on in the literature (Lancione et al., 2022) and will need to be continually addressed to promote large-scale, harmonized studies of this novel MRI metric.Although still in early, innovative stages, our study contributes to supporting the validity and wider effort understanding the utility of MRI QSM for future studies.Importantly, approximately 85% of our cohort identified as non-Hispanic White, which reinforces the fieldwide need to accurately represent communities and racial groups that are insufficiently included in our research populations.We aim to replicate our findings in a larger, increasingly more ethnically and racially diverse group of older adults to determine the generalizability of observed relationships.

Conclusion
As a modifiable lifestyle factor, PA continues to be an empirically supported target for primary cognitive health prevention in clinically normal older adults.Given that these are the first known data to report on relationships among iron deposition, PA, and memory, this study innovatively utilizes QSM as an in vivo method for exploring novel pathways that link lifestyle behaviors to cognitive health in humans.Clinically, our findings may offer support for the utility of QSM for the risk stratification of exercise interventions.Elevated iron deposition confers risk for cognitive decline and neurodegenerative disease (Ravanfar et al., 2021).By encouraging PA, we aim to attenuate the clinical manifestation of accumulating brain iron deposition in aging adults.Beyond the scope of this study, future multimodal longitudinal randomized controlled trials that incorporate PA interventions, fluid biomarkers, and QSM will enable us to temporally capture in vivo neurobiological changes that directly inform PA pathways and their specific mechanistic effects on human cognition.

Table 2
Bilateral hippocampal model results