Age and sex differences in the associations of pulse pressure with white matter and subcortical microstructure

Abstract

Midlife vascular disease increases risk for dementia, and effects of vascular dysfunction on brain health differ between men and women. Elevated pulse pressure, a surrogate for arterial stiffness, contributes to cerebrovascular pathology and white matter damage that may advance cognitive aging; however, it remains unclear how associations between pulse pressure and neural integrity differ by sex and age. This study used restriction spectrum imaging (RSI) to examine associations between pulse pressure and brain microstructure in community-dwelling women (N=88) and men (N=55), aged 56–97 (mean 76.3) years. Restricted isotropic (presumed intracellular), hindered isotropic (presumed extracellular), neurite density, and free water diffusion were computed in white matter tracts and subcortical regions. After adjustment for age and sex, higher pulse pressure corelated with lower restricted isotropic diffusion in global white matter, with more pronounced associations in parahippocampal cingulum, as well as in thalamus and hippocampus. Subgroup analyses demonstrated stronger correlations between pulse pressure and restricted isotropic diffusion in association fibers for participants ≤75 years than for older participants, with stronger effects for women than men of this age group. Microstructure in parahippocampal cingulum and thalamus differed by pulse pressure level regardless of antihypertensive treatment. Increased pulse pressure may lead to widespread injury to white matter and subcortical structures, with greatest vulnerability for women in late middle to early older age. RSI could be useful for monitoring microstructural changes indicative of neuronal loss or shrinkage, demyelination or inflammation that accompany age-related cerebrovascular dysfunction.

Graphical Abstract

Introduction

Increased arterial stiffness predicts future cardiovascular events¹ and promotes neurovascular dysfunction that may accelerate cognitive decline with age.² There is mounting evidence that cerebrovascular injury accumulates across the lifespan, with changes that emerge in midlife posing a stronger risk for dementia than those emerging in later- life.³ This age dependence suggests that deleterious consequences of cerebrovascular dysfunction materialize early and guide trajectories of neurobiological aging. White matter is exceptionally vulnerable to hypoperfusion and microvascular damage, displaying characteristic lesions consequent to vascular pathology.⁴ However, subtle microscopic changes reflecting demyelination or axonal degeneration may present in the absence of macroscopic damage and serve as early warning signs of latent vascular pathology. Identifying the most sensitive markers of vascular-related cytoarchitectural injury and disentangling their impact across critical life periods may open a pivotal window for timely intervention to preserve brain health in aging.

Arterial stiffness can be estimated by pulse wave velocity (PWV) or pulse pressure (PP), the difference between systolic and diastolic blood pressures. Age-related arterial stiffening impairs vessel compliance, elevating pulsatile forces along the brain’s delicate microvasculature and leading to hypoperfusion, impaired cerebral autoregulation, neuroinflammation and blood-brain barrier leakage.⁵ These pathophysiological consequences in turn promote small vessel damage and neuronal injury that often culminate in lacunar infarcts, microbleeds and white matter hyperintensities (WMH).⁶ Whereas WMHs index demyelination, axon loss or gliosis from longstanding hypoxia,⁷ diffusion imaging may be better positioned to examine microstructural changes associated with early or mild vascular dysfunction preceding gross lesions.⁸ Diffusion tensor imaging (DTI) studies have consistently reported reduced white matter fractional anisotropy (FA) and increased mean diffusivity (MD) with hypertension.^9–12 Fewer studies have examined relationships between arterial stiffness and white matter microstructure, reporting inconclusive support for reduced FA and increased diffusivity^13–15 or no association beyond that for WMH,¹⁶ with PP or PWV. There has been minimal examination of subcortical regions, despite vulnerability of the basal ganglia to dilated perivascular spaces, lacunes and white matter hyperintensities related to cerebral small vessel disease,¹⁷ and reports that hypertensive individuals demonstrate hippocampal atrophy, hypoperfusion and hypometabolism.^18–20

Midlife vascular risk is more strongly linked to subsequent dementia than is later-life vascular risk.³ Though less well-studied than cognitive outcomes, vascular-related brain changes are similarly less consistent in older than younger age. Elevated PP and PWV have been associated with WMH and reduced FA in cohorts with mean ages in their sixties,^{15, 21} but minimal associations have been reported among individuals two decades older.¹⁶ Direct comparison across ages has revealed stronger correlations of PP with WMH and diffusivity among those over age 35¹⁴ or 65²¹ than in younger groups. However, because few studies have evaluated associations in the oldest old (over age 80), it remains uncertain how white matter vulnerability to vascular dysfunction evolves during the final decades of life.

Vascular dysfunction emergent in midlife may manifest as more severe brain injury in older age because of prolonged pathology, or conversely, may encourage timely intervention that mitigates early cerebrovascular complications. Whereas some studies suggest that antihypertensive treatment counters white matter damage,^{12, 22, 23} others observed no protective effect^{9, 24} or worse²⁵ white matter integrity for treated individuals. Thus, further research is warranted to unravel the complex interplay between the severity and treatment of vascular pathology.

There is mounting evidence that vascular risk factors differ by sex in prevalence and neurobiological impact.²⁶ During later life, rates of stroke, hypertension and atrial fibrillation are higher in women,²⁷ potentially compounded by lower brain or cognitive reserve in women that may exacerbate neural insult from vascular pathology.²⁸ Indeed, hypertension, obesity and diabetes are differentially associated with dementia for men and women, and predict more rapid cortical atrophy for cognitively healthy older women.^29–31 Understanding sex differences in how arterial stiffness impacts cytoarchitecture will help to clarify this differential susceptibility to vascular-related brain injury.

Thus, there is outstanding need to more thoroughly characterize effects of arterial stiffness on white matter and subcortical microstructural integrity in aging men and women and to assess whether such changes are modifiable with therapy. Multicompartment diffusion MRI models are more sensitive to subtle cytoarchitectural changes due to cell damage or inflammation than conventional DTI, which is unable to resolve within-voxel complexities or dissociate diffusion among distinct cellular spaces. Restriction spectrum imaging (RSI) parses diffusion among intracellular, extracellular and cerebrospinal fluid (CSF) compartments,³² affording better characterization of tissue cytoarchitecture for tumor detection ³³ and tracking changes associated with epilepsy^{34, 35} and Alzheimer’s disease,³⁶ than does DTI. Using RSI, we previously observed marked microstructural brain changes with age that are stronger for women³⁷ and may mediate cognitive decline (Reas et al., submitted), though it remains unclear how these cytoarchitectural properties are modified by age-related arterial stiffening.

Here, we used RSI to examine cross-sectional associations of PP with white matter fiber and subcortical microstructure in a sample of cognitively healthy older adults with ages spanning from midlife to the oldest old. PP-microstructure correlations were compared between younger and older age groups to assess how vascular-related brain injury varies with age, and between men and women to identify sex differences in vulnerability to arterial stiffness. Finally, we examined whether microstructure in those with normal PP in conjunction with antihypertensive treatment resembled that of individuals with elevated PP.

Methods

Anonymized demographic and health data are available at knit.ucsd.edu/ranchobernardostudy. Imaging data are available upon request.

Participants

Participants were 154 community-dwelling southern California residents enrolled in the Rancho Bernardo Study (RBS) of Healthy Aging³⁸ who completed a research visit in 2014–2016 and underwent an MRI. Exclusion criteria included history of head injury, stroke, neurological disease, treatment for an alcohol use disorder, or safety contraindication for MRI. After excluding six individuals due to poor data quality, four due to missing blood pressure information, and one due to severe white matter disease, the final sample included 55 men and 88 women (mean age 76.3±7.5; range 56–97 years).

Study procedures were approved by the University of California, San Diego Human Research Protections Program Board and participants provided informed written consent prior to participation.

Pulse pressure measurement

Blood pressure was measured in seated, resting participants by a nurse certified in the Hypertension Detection and Follow-up Program protocol, using a regularly calibrated standard mercury sphygmomanometer. The mean of two readings, taken five minutes apart, was used for analysis. Pulse pressure (PP) was computed as the difference between systolic blood pressure (SBP) and diastolic blood pressure (DBP), and PP >60 mm Hg was deemed high based on clinical guidelines for cardiovascular risk.⁴⁶ Participants were considered hypertensive if they had an average SBP≥140, DBP≥90, were taking antihypertensive medication, or reported a physician diagnosis of hypertension.

Cognitive, health, and lifestyle assessment

The Modified Mini Mental Status Examination (3MS), a screening tool that assesses multiple cognitive domains, was administered by a trained examiner in a quiet room. Education level was assessed at enrollment and converted to years of education. Height and weight were measured with the participant wearing light clothing, without shoes, and used to compute body mass index (BMI, kg/m²). History of smoking (never versus former; there were no current smokers) and diabetes (yes/no), along with current exercise (three or more times per week, yes/no), alcohol consumption (non-drinker/drinker), and antihypertensive medication use (yes/no), were obtained from standard questionnaires.

Imaging data acquisition

Imaging data were acquired on a 3.0 Tesla Discovery 750 scanner (GE Healthcare, Milwaukee, WI, USA) with an eight-channel phased array head coil at the University of California, San Diego Center for Functional MRI. The MRI sequence included a three-plane localizer; a sagittal 3D fast spoiled gradient echo T₁-weighted volume optimized for maximum gray/white matter contrast (TE=3.2 ms, TR=8.1 ms, inversion time=600 ms, flip angle=8°, FOV=256×256 mm, matrix=256×192, slice thickness=1.2 mm, resampled to a resolution of 1×1×1.2 mm, scan time 8:27); and an axial 2D single-shot pulsed-field gradient spin-echo echo-planar diffusion-weighted sequence (45-directions, b-values=0, 500, 1500, 4000 s/mm2, one b=0 volume and 15 gradient directions for each non-zero b-value; TE=80.6 ms, TR=7 s, FOV=240×240 mm, matrix=96×96, slice thickness=2.5 mm, resampled to a resolution of 1.875×1.875×2.5 mm, scan time 6:34).

Data processing

Image processing steps are outlined in Figure S1. MRI data were processed using an automated pipeline combining FreeSurfer (http://surfer.nmr.mgh.harvard.edu) with tools developed in-house, as previously described.³⁹ Diffusion MRI data were corrected for distortions^40–42 and visually inspected for artifacts. Participants with images containing uncorrectable artifacts (N=6) were excluded from analysis. Gray matter, white matter, and CSF boundaries were identified on T₁-weighted structural images using FreeSurfer. Diffusion MRI data were automatically registered to T₁ images.⁴³

Computation of RSI metrics

RSI metrics of interest included restricted isotropic (RI), neurite density (ND), hindered isotropic (HI), and isotropic free water (IF) diffusion.^{32, 39} RI, thought to correspond with intracellular diffusion, was computed as the 0^th spherical harmonic of the restricted compartment. ND combined measures of restricted oriented and crossing fibers, computed as the 2^nd and 4^th spherical harmonics of the restricted compartment, respectively, into a measure corresponding with the density of axons or dendrites. HI was computed as the 0^th spherical harmonic of the hindered compartment, which presumably corresponds with diffusion within the extracellular space or large cell bodies. IF estimates the CSF fraction.

RSI parameters were measured in five subcortical regions (thalamus, caudate, putamen, hippocampus, and amygdala), automatically segmented according to a subcortical atlas.⁴⁴ HI was not examined in fiber tracts because white matter microstructure is poorly characterized by the hindered fraction,³² but all other RSI metrics were computed in fifteen white matter fiber tracts derived from a probabilistic atlas-based fiber tract segmentation.⁴⁵ Voxels containing primarily gray matter or CSF were excluded from white matter tracts.⁴⁴ Fiber tracts demonstrating significant associations with PP are illustrated in Figure S2. Global white matter RSI measures were calculated as the mean across all fiber tracts. To minimize partial volume effects for cortical gray matter surface-based analyses, RSI metrics were sampled with linear interpolation from 0.8–2.0 mm from the gray/white matter boundary along the normal to the cortical ribbon and combined using a weighted average based on the proportion of gray matter in each voxel. RSI cortical surface maps were registered to common space (fsaverage) and smoothed with a FWHM 10 mm kernel.

Statistical analysis

Participants were classified as having normal or high PP (>60 mm Hg).⁴⁶ Age, health and lifestyle variables were compared between PP groups using independent-samples t-tests (continuous variables) or chi-squared tests (categorical variables). 3MS scores, adjusted for age, sex and education, and global RSI measures, adjusted for age and sex, were compared between groups using univariate ANOVA. Correlations among health and demographic variables were computed using Pearson’s correlations.

To examine associations between PP and microstructure, partial correlations, adjusted for age and sex, were computed between PP and RSI metrics across global white matter, within regional fiber tracts, and within subcortical regions. For fibers demonstrating significant correlations between PP and RSI, partial correlations were repeated with additional adjustment for the respective global RSI metric.

Though our primary focus was on white matter and subcortical microstructure, exploratory analyses examined associations between PP and cortical gray matter microstructure. Vertex-wise general linear models (GLMs), with a regressor of PP covaried for age and sex, were computed for each RSI metric along the cortical surface.

To evaluate whether associations between PP and microstructure differed by age, participants were dichotomized into those 75 years old or less (N=63) and older than 75 years (N=80). Partial correlations of PP with white matter and subcortical microstructure were repeated, as described above, stratified by age group. For any measure showing a significant correlation, correlation strengths were compared between age groups by converting correlation coefficients to normalized correlations (z) (Fisher’s r-to-z transformation) and computing the confidence interval of the difference between correlations.

To examine sex differences in associations between PP and microstructure, partial correlations between PP and RSI metrics were repeated, stratified by sex, and ANOVAs comparing RSI metrics by PP level were conducted with a sex by PP interaction term. To assess whether sex differences in PP-microstructure associations differed by age, correlations were repeated, stratified by sex and age group. Sex differences in correlation strengths were assessed using Fisher’s r-to-z transformation.

Antihypertensive treatment may mitigate microstructural damage by lessening vascular burden or conversely, may be more prevalent in those with chronic vascular dysfunction who may manifest more severe, longstanding brain injury. Therefore, to assess how treatment modifies PP-related differences in microstructure, participants were classified into three groups based on PP and antihypertensive treatment: untreated with normal PP (N=56), treated with normal PP (N=33), and high PP (N=54). White matter and subcortical RSI measures were compared among groups using univariate ANOVA, adjusted for age and sex. Post-hoc pair-wise comparisons (Bonferroni corrected for multiple comparisons, p<0.05) were computed for RSI measures demonstrating significant group effects.

Cortical surface analyses were performed in FreeSurfer version 6.0 and all other analyses were conducted in SPSS version 27.0 (IBM Corp, Armonk, NY, USA) and significance was set to p<0.05. Significance for regional analyses were adjusted for multiple comparisons across 15 fiber tracts (p<0.0033) or five subcortical regions (p<0.01). Exploratory analyses for cortical surface GLMs were examined both with false discovery rate (FDR) correction (p<0.05) and at an uncorrected threshold (p<0.01).

Results

Participants

Mean PP was 56.0±15.0 mm Hg (range 29–99 mm Hg), and 38% of participants (N=54) met criteria for high PP (>60 mm Hg). Participant health and demographic characteristics by PP level are presented in Table 1. Those with high PP were older, had higher SBP, and were more likely to be taking antihypertensive medications than those with normal PP (p<0.001). PP groups did not differ by sex, education, DBP, exercise, smoking, alcohol consumption, BMI, diabetes or 3MS scores.

Table 1.

Participant characteristics and mean RSI measures (mean±SD unless otherwise noted) for those with normal or high pulse pressure.

Measure	PP ≤60 mm Hg (N=89)	PP >60 mm Hg (N=54)	Group effect
Age (years) [Range]	74.2±7.1 [56–97]	79.9±7.0 [63–96]	p<0.001
Sex (% women)	60	65	p=0.53
Systolic blood pressure	`117±12	144±15	p<0.001
Diastolic blood pressure	71±9	72±11	p=0.55
Pulse pressure	46.5±7.9	71.6±9.9	p<0.001
Antihypertensives (%)	48	93	p<0.001
Education (years)	14.9±2.2	14.8±1.8	p=0.75
Exercise (% 3x/week)	79	72	p=0.38
Body mass index (kg/m2) *	25.7±3.8	26.4±4.2	p=0.27
Smoking (% ever)	39	44	p=0.55
Alcohol (% drinker)	87	83	p=0.60
Diabetes (%)	12	9	p=0.57
3MS †	95.0±4.1	95.1±4.5	p=0.90
Fiber RI ‡	0.425±0.023	0.416±0.023	p=0.02
Fiber ND	0.601±0.022	0.599±0.024	p=0.55
Fiber IF	0.250±0.028	0.251±0.033	p=0.81

^*Body mass index is adjusted for sex.

^†3MS (modified mini-mental) scores are adjusted for age, sex and education.

^‡RSI scores are adjusted age and sex.

IF, isotropic free water; ND, neurite density; PP, pulse pressure; RI, restricted isotropic

Correlations of PP with white matter and subcortical microstructure

As shown in Figure 1, higher PP correlated with lower global fiber RI (r=−0.22, p=0.009; adjusted for age and sex). Regionally, higher PP correlated with lower RI in the parahippocampal cingulum (r=−0.33, p<0.001), uncinate (r=−0.26, p=0.002), and IFO (r=−0.25, p=0.003) (see Figure S2 for location of fiber tracts). The correlation for parahippocampal cingulum remained after adjustment for global RI (r=−0.26, p=0.002). In subcortical regions, higher PP correlated with lower RI in thalamus (r=−0.25, p=0.003) and hippocampus (r=−0.26, p=0.002). PP did not correlate with ND or IF in any fiber tract or subcortical region examined (p>0.05).

Figure 1. Correlations of PP with white matter and subcortical microstructure.

Partial correlations are shown between PP and restricted isotropic (RI) diffusion across all fiber tracts and within subcortical regions. RSI measures are standardized residuals (adjusted for age and sex).

Associations between PP and cortical gray matter microstructure

Cortical gray matter microstructure did not significantly correlate with PP (p>0.05, FDR corrected; adjusted for age and sex). Without FDR correction (p<0.01), scattered correlations were present, with moderate correlations between higher PP and higher cortical HI (Figure S3).

Age differences in associations between PP and microstructure

Among participants aged 75 years or less (N=63), higher PP correlated (adjusted for age and sex) with lower global fiber RI (r=−0.35, p=0.005) and regionally, with lower RI in parahippocampal cingulum, uncinate, inferior longitudinal fasciculus (ILF), IFO, and forceps minor (p<0.0033) (tracts are shown in Figure S2). Within subcortical regions, higher PP correlated with lower RI in hippocampus r=−0.34, p=0.008) and amygdala (r=−0.33, p=0.01) for younger-old participants. Correlations were stronger for younger-old than older-old individuals for RI in the uncinate (z=2.77, p=0.006), ILF (z=2.10, p=0.04), IFO (z=2.62, p=0.009), and forceps minor (z=2.41, p=0.02) (Figure 2). Among individuals older than 75 years (N=80), higher PP correlated with higher thalamus HI (r=0.33, p=0.004), but this correlation did not significantly differ (p=0.25) from that for young-old participants (r=0.14, p=0.27).

Figure 2. Correlations between PP and microstructure by age.

Partial correlations are shown between PP and restricted isotropic (RI) diffusion for 63 younger-old (≤75 years) and 80 older-old (>75 years) participants for measures demonstrating significant age differences in correlation strength. RSI measures are standardized residuals (adjusted for age and sex).

Sex differences in associations between PP and microstructure

Across all subjects, there were neither significant sex differences in correlations between PP and microstructure, nor significant interactions between sex and PP level (normal versus high) on microstructure. To examine sex differences by age group, men and women were compared when stratified by age (see Table S1 for participant characteristics by sex and age group). Among those ≤75 years, women showed stronger correlations than men of PP with global white matter RI (women: r=−0.53, p<0.001; men: r=0.00, p=1.00; z=2.01, p=0.04), and with RI in ILF (women: r=−0.58, p<0.001; men: r=−0.02, p=0.95; z=2.20, p=0.03) and superior longitudinal fasciculus (women: r=−0.46, p=0.002; men: r=0.19, p=0.42; z=2.38, p=0.02) (Figure 3; Figure S2). Correlations did not differ by sex among those older than 75 years.

Figure 3. Correlations between PP and microstructure by sex for younger-old participants.

Partial correlations are shown between PP and restricted isotropic (RI) diffusion for younger-old (≤75 years) women and men for measures demonstrating significant sex differences in correlation strength. RSI measures are standardized residuals (adjusted for age).

Differences in microstructure by PP and antihypertensive treatment

RSI measures (adjusted for age and sex) were compared across untreated individuals with normal PP (PP≤60, N=56), treated individuals with normal PP (N=33), and individuals with high PP (PP>60, N=54). There were no group differences in global white matter microstructure (p>0.05). Regionally, groups differed for parahippocampal cingulum RI (F(2,138)=7.57, p<0.001), thalamus RI (F(2,134)=5.57, p=0.005), and thalamus HI (F(2,133)=5.35, p=0.006) (Figure 4). Pairwise comparisons (Bonferroni corrected for comparison across three groups) revealed lower RI and higher HI in thalamus for those with high PP than those with normal PP, regardless of treatment (p<0.05). Parahippocampal cingulum RI was lower for those with high PP than those with normal PP who were untreated (p<0.001). No measure showed a difference between treated and untreated participants with normal PP.

Figure 4. Differences in microstructure by PP and antihypertensive treatment.

RSI measures demonstrating significant group differences are shown for untreated participants with normal PP (≤60 mm Hg), treated participants with normal PP, and participants with high PP (>60 mm Hg). RSI measures are standardized residuals (adjusted for age and sex). HI, hindered isotropic; RI, restricted isotropic

Discussion

In this study, altered restricted and hindered isotropic diffusion in white matter tracts and subcortical regions, but not in cortical gray matter, were associated with elevated PP in non-demented, community-dwelling older adults. These results support prior observations of white matter compromise with arterial stiffness and extend these findings to demonstrate cytoarchitectural and topographic specificity within association fibers and subcortical structures. Critically, these results offer novel evidence that injurious associations of arterial stiffness on microstructure vary by age and sex, with more pronounced correlations among those aged 56–75 years than for older individuals, particularly among women of this age range. Finally, our findings suggest that preventing arterial stiffening via effective antihypertensive treatment may help to mitigate neuronal injury.

Across four decades of age spanning middle to late life, higher PP correlated with globally reduced white matter RI, with particularly strong associations in the parahippocampal cingulum. Age- and sex-specific analyses revealed additional correlations in association and commissural fibers, including the uncinate, ILF, IFO, SLF, and forceps minor. Previous studies have found mostly global white matter changes, with more limited localized effects, related to surrogates of arterial stiffness.^{14, 47} However, the regional specificity observed here supports a larger body of research reporting localized correlations between hypertension and altered diffusivity in association tracts including those identified in here.^{9, 10, 24, 25} Furthermore, higher PP was also associated with reduced RI in the hippocampus and thalamus and increased thalamus HI. These results present compelling evidence for cytoarchitectural changes related to arterial stiffening in subcortical regions that also manifest infarcts, lacunes and WMHs with small vessel disease.¹⁷

PP better predicted white matter RI for individuals from middle to early older age than for those over age 75, consistent with consensus that midlife presents a crucial window wherein vascular dysfunction substantially elevates risk for dementia. Thus, neural injury may be most profound during midlife when vascular changes begin to develop, which may contribute to pathophysiological changes underlying cognitive decline. A diminishing effect of PP on white matter microstructure with age aligns with prior studies reporting deleterious effects of arterial stiffness on white matter in cohorts with mean ages below 70 years^{13–15, 21} but weaker effects for older cohorts.¹⁶ By directly comparing age groups, the present study avoids confounding due to cohort or methodological differences between studies. However, the mechanisms by which the strength of vascular-related brain injury dissipates with age are unclear. Furthermore, selection bias may obscure associations in older participants due to exclusion of individuals with confounding conditions or to differences in survival related to PP. Cross-sectional studies like ours preclude evaluation of intra-individual microstructural remodeling that may accompany dynamic vascular changes, warranting further longitudinal assessment.

Despite mounting evidence that sex and vascular dysfunction both significantly shape neurobiological aging trajectories, their interactive effects on brain health have been inadequately examined. Most studies of microstructural changes related to arterial stiffness have covaried for sex rather than evaluating differential risk by sex.^{13–15, 21, 47} In line with reports that correlations between PWV and white matter FA do not differ by sex in the very old,¹⁶ we also observed no sex differences among individuals over age 75. However, among younger participants, stronger associations for women than men emerged between PP and fiber RI. These results expand upon reports that hypertension predicts atrophy for women but not men,²⁹ to suggest that arterial stiffness is also related to risk for white matter damage in women. Furthermore, this suggests that susceptibility to vascular-related neural injury for women is age-dependent, with the transition from middle to older age presenting a critical period for preventing cerebrovascular damage.

Among individuals with low PP, microstructure was comparable between individuals treated and untreated for hypertension. This suggests that variations in current vascular burden, which may be attenuated with effective therapy, better predicts neural integrity than history of vascular risk. These findings align with reports that WMH and FA are comparable between normotensives and individuals with controlled hypertension^{12, 23} and further indicate that effective treatment may mitigate both white matter and subcortical damage by preserving arterial compliance. However, this conclusion contradicts other observations that FA or MD differ between hypertensive and normotensive participants but are not modified by blood pressure control.^{9, 24} Different antihypertensive classes have varying effects on arterial stiffness,⁴⁸ and because arterial stiffness may precede blood pressure elevations,⁴⁹ targeting PP reductions rather than blood pressure may more effectively prevent neurovascular damage.

Lower RI, a measure of intracellular diffusion, in white matter, thalamus and hippocampus may reflect reduced axon density, demyelination, cell shrinkage or dystrophy. Increased thalamus HI may correspond with expansion of the extracellular space or fluid accumulation, perhaps related to cell shrinkage or inflammation. Minimal associations were observed in cortical gray matter, highlighting the specificity of vascular-related microstructural changes to white matter and subcortical regions. Broadly, these results align with evidence of reduced FA and increased diffusivity in white matter, with PP or PWV.⁵ However, variations in blood flow and pressure induce intracranial pulsatility,⁵⁰ which increases with vascular pressure and could theoretically influence diffusion measures. Such effects might manifest more strongly near large vessels experiencing more severe pulsative forces. Potential effects of vascular pulsatility on dynamic imaging parameters remain speculative and warrant investigation. We observed no associations for neurite density (restricted oriented diffusion) or free water, the closest RSI analogues to FA and mean diffusivity, respectively. Such discrepancies may be attributed to the greater information afforded by RSI, which accounts for crossing fibers that can dilute estimates of mean voxel anisotropy, and parcels isotropic diffusion into free water, restricted and hindered compartments. By adjusting models for age, this study isolates RI and HI as vascular injury markers from widespread age-related changes in isotropic, anisotropic and free water metrics throughout gray and white matter,³⁷ suggesting that age- and vascular-related cytoarchitectural damage are dissociable using multicompartment diffusion imaging models.

Because RSI approximates cell architecture, further histological study is needed to clarify the biological substrates of the differences reported here. Although this sample was free of dementia, concomitant preclinical neuropathology could contribute to microstructural alterations, potentially modifying age differences due to heavier neuropathological burden in the oldest old. Because of the small number of participants in the sex -and age-stratified analyses, further study is needed to replicate these findings. Finally, PP indirectly estimates arterial stiffness, and because it depends upon blood pressure, dissociating effects of blood pressure and arterial stiffness was not possible here. However, PP can be calculated using inexpensive, noninvasive methods relative to PWV, allowing individuals to conveniently monitor changes from the clinic or home.

Perspectives

In summary, elevated PP among community-dwelling older adults is related to white matter and subcortical markers of neural cytoarchitecture consistent with reduced intracellular and expanded extracellular spaces. Vascular-related microstructural remodeling may become particularly pronounced for women in midlife and begin to taper around the eight decade of life. However, lowering PP via antihypertensive therapy may minimize vascular burden and mitigate brain injury. RSI may be valuable for monitoring subtle microstructural brain differences associated with incipient vascular dysfunction, which could enable timely intervention to slow progressive cerebrovascular injury.

Novelty and Significance.

What is new?

• Using restriction spectrum imaging (RSI), we found that increased pulse pressure is associated with altered brain microstructure in white matter and subcortical regions, which may reflect cell or axon loss, demyelination or inflammation.

• Magnitude of pulse pressure-related microstructural injury differs by age and sex, with greater associations for women in early older age than for men or for older individuals.

• Microstructural differences with pulse pressure were independent of antihypertensive medication use.

What is relevant?

• Subtle cellular injury associated with elevated pulse pressure may be most pronounced for women immediately following midlife.

• RSI may be useful for monitoring microstructural brain changes indicative of latent vascular pathology, helping to guide approaches to preserve cerebrovascular health into old age.

Summary:

Elevated pulse pressure may induce microstructural injury to white matter and subcortical regions, with the most profound effects for women in late middle to early older age. Vascular-related microstructural neural damage can be identified and tracked using RSI.