Aerobic exercise improves hippocampal blood flow for hypertensive Apolipoprotein E4 carriers

Abstract

Cerebrovascular dysfunction likely contributes causally to Alzheimer’s disease (AD). The strongest genetic risk factor for late-onset AD, Apolipoprotein E4 (APOE4), may act synergistically with vascular risk to cause dementia. Therefore, interventions that improve vascular health, such as exercise, may be particularly beneficial for APOE4 carriers. We assigned cognitively normal adults (65–87 years) to an aerobic exercise intervention or education only. Arterial spin labeling MRI measured hippocampal blood flow (HBF) before and after the 52-week intervention. We selected participants with hypertension at enrollment (n = 44). For APOE4 carriers, change in HBF (ΔHBF) was significantly (p = 0.006) higher for participants in the exercise intervention (4.09 mL/100g/min) than the control group (−2.08 mL/100g/min). There was no difference in ΔHBF between the control (−0.32 mL/100g/min) and exercise (−0.54 mL/100g/min) groups for non-carriers (p = 0.918). Additionally, a multiple regression showed an interaction between change in systolic blood pressure (ΔSBP) and APOE4 carrier status on ΔHBF (p = 0.035), with reductions in SBP increasing HBF for APOE4 carriers only. Aerobic exercise improved HBF for hypertensive APOE4 carriers only. Additionally, only APOE4 carriers exhibited an inverse relationship between ΔSBP and ΔHBF. This suggests exercise interventions, particularly those that lower SBP, may be beneficial for individuals at highest genetic risk of AD.

ClinicalTrials.gov Identifier: NCT02000583

Introduction

Evidence increasingly points to an early, primary role of cerebrovascular dysfunction in the pathogenesis of late-onset Alzheimer’s disease (AD).^1–4 Considering the well-established benefits of aerobic exercise on vascular health,⁵ exercise may act mechanistically through improvements in vascular function to reduce dementia risk. Indeed, intervention trials have shown exercise may improve cognitive function at least partially through increasing cerebral blood flow (CBF),^6–9 particularly in the hippocampus (HBF).^10–12

The strongest known genetic risk factor for AD, the APOE4 allele, may act synergistically with poor vascular health to increase dementia risk.^13–17 This suggests interventions that improve systemic vascular health may be particularly beneficial for APOE4 carriers.¹⁸ Some studies have shown APOE4 carriers benefit more from exercise,^19–22 but others have suggested less improvement in APOE4 carriers,^23,24 meriting further exploration.

We performed an analysis of a secondary outcome from a randomized controlled trial in which cognitively normal older adults were assigned to a 52-week aerobic exercise intervention or to education only.²⁵ For the present analysis, we selected only participants with hypertensive blood pressure at the time of enrollment, defined as systolic blood pressure (SBP) ≥130 mmHg or diastolic blood pressure (DBP) ≥80 mmHg.²⁶ We hypothesized that 1) the exercise intervention would be more effective in increasing HBF in APOE4 carriers than non-carriers, and 2) reductions in SBP would have a greater impact on improving HBF for APOE4 carriers than for non-carriers.

Material and methods

Study design

The Alzheimer’s Prevention through Exercise study (APEx) was a 52-week clinical trial of aerobic exercise in cognitively normal older adults. The primary outcome was change in β-amyloid deposition, and these results have been published previously.²⁵ The present study is an analysis of the secondary outcome of regional arterial spin labeling (ASL) MRI data, specifically HBF, for the participants in the APEx clinical trial (NCT02000583).

Participants

Participants were required to meet the following inclusion criteria: 65 years and older, sedentary or underactive as defined by the Telephone Assessment of Physical Activity, on stable medications for at least 30 days, willingness to undergo an 18 F-AV45 PET scan for cerebral β-amyloid load and learn the result (elevated or non-elevated), willingness to perform prescribed exercise (or not) for 52 weeks at a community fitness center, and ability to complete graded maximal exercise testing with a respiratory exchange ratio >=1.0. Exclusion criteria included insulin-dependence, significant hearing or vision problems, clinically evident stroke, cancer in the previous 5 years (except for localized skin or cervical carcinomas or prostate cancer), change in blood pressure medication within the last 30 days, or recent history (<2 years) of major cardiorespiratory, musculoskeletal or neuropsychiatric impairment. During in-person screening, a clinician of the University of Kansas Alzheimer’s Disease Center performed a clinical assessment that included a Clinical Dementia Rating, the Uniform Data Set neuropsychiatric battery, and other supplementary tests.

Neuroimaging assessments

Individuals who consented to screening underwent florbetapir 18 F-AV45 (370 MBq) PET scans. β-amyloid status was disclosed to all participants.²⁷ We enrolled those participants with cortical-to-cerebellar β-amyloid burden greater than 1 because these participants may have accelerated β-amyloid deposition and memory decline.²⁸ At baseline, enrollees had a T1-weighted MRI of the brain (3 T Siemens Skyra scanner; MP-RAGE 1 × 1 × 1.2 mm voxels, TR = 2300 ms, TE = 2.98 ms, TI = 900 ms, FOV 256 × 256×mm, 9° flip angle; Pulsed ASL single-shot EPI 3.8 × 3.8 × 4.0 mm,TR = 3400 ms, TE = 13 ms, TI = 700 ms, FOV 240 × 240×mm, 90° flip angle) with regional volumes parcellated and extracted using the CAT12 toolbox (http://www.neuro.uni-jena.de/cat/) and the Automatic Anatomic Labeling atlas. ASL-MRI data were processed using the ASLTbx for SPM12.²⁹ All neuroimaging assessments were repeated at 52 weeks.

Physiological assessments

At the baseline study visit, the participant sat at rest for 5 minutes before BP was measured twice with one minute of rest between measurements (Axia TRIA Touch Screen Patient Monitor, Association for the Advancement of Medical Instrumentation/American National Standards Institute performance standards SP10:2002). We averaged the two resting SBP and two resting DBP to determine one average baseline SBP/DBP, which was used for grouping into hypertensive and normotensive categories based on the most recent guidelines published by the American College of Cardiology and American Heart Association (ACC/AHA).²⁶ Before beginning the study, participants performed graded maximal exercise testing on a treadmill to maximal capacity or volitional termination to quantify cardiorespiratory fitness (VO₂max). The BP measurement and graded maximal exercise test were repeated at 52 weeks.

Cognitive assessments

A trained psychometrist performed a comprehensive cognitive test battery at baseline and 52 weeks, employing validated, alternate versions of tests every other visit. We created composite scores for three cognitive domains (executive function, verbal memory, visuospatial processing) using Confirmatory Factor Analysis. Scores were standardized to baseline so subsequent scores could be interpreted relative to baseline. The executive function composite score was made up of verbal fluency, Trailmaking Test B, Digit Symbol Substitution test, and the interference portion of the Stroop test. The verbal memory composite score was made up of the Logical Memory Test immediate and delayed, and the Selective Reminding Test. The visuospatial composite score was made up of scores from Block Design, space relations, the paper folding test, hidden pictures, and identical pictures. Missing data in the factor analysis were accounted for using full information maximum likelihood algorithm.

Intervention

Participants were randomized in a 2:1 ratio to either 150 minutes per week of supported moderate intensity aerobic exercise or standard of care education. The education control group was provided standard exercise public health information but was otherwise not supported nor prohibited from exercise. For those randomized to the aerobic exercise group, the intervention was conducted at their nearest study-certified exercise facility with the support of certified personal trainers. The intervention group was asked to refrain from changing their regular physical activities other than those prescribed by the study team. Participants exercised 3–5 days a week at an intensity that began at 40–55% of Heart Rate Reserve (% of the difference between maximal and resting) and was increased by 10% every 3 months.

APOE genotype determination

Whole blood was collected and stored at −80 C until genetic analyses could be conducted. To determine APOE genotype, frozen whole blood was assessed using a Taqman single nucleotide polymorphism (SNP) allelic discrimination assay (ThermoFisher). APOE4, APOE3, and APOE2 alleles were distinguished using Taqman probes to the two APOE-defining SNPs, rs429358 (C_3084793_20) and rs7412 (C_904973_10). The term “APOE4 carrier” was used to describe the presence of 1 or 2 APOE4 alleles. Since APOE2 is associated with reduced AD risk, all APOE2 carriers were excluded from the analysis (whether homozygous or paired with a different APOE allele).

Standard protocol approvals, registrations, and patient consents

All study (ClinicalTrials.gov, NCT02000583; trial active between 11/1/2013–11/6/2019) procedures were approved by the University of Kansas Institutional Review Board (HSC #13376) and complied with the World Medical Association Declaration of Helsinki. Written informed consent was obtained from all participants.

Statistical approach

All statistical analyses were performed using SPSS Statistics (IBM). Normality was assessed before each analysis by Shapiro-Wilk’s test (p > 0.05) and Q-Q Plot. Baseline group differences were assessed by independent t-test, chi-square test for homogeneity, Mann Whitney U, Kruskal-Wallis H test, or one-way ANOVA, as appropriate. A multiple linear regression further characterized the relationship between SBP, DBP and baseline HBF. A two-way ANCOVA was conducted to examine the effects of APOE4 carrier status and Treatment Group (exercise or control) on ΔHBF, after controlling for age and sex. The significant interaction term was followed up by an analysis of simple main effects using a Bonferroni adjustment (p < 0.025). The same two-way ANCOVA analysis was performed with outcome measure percent ΔHBF (%ΔHBF = (HBF at 52 weeks – HBF at baseline)/HBF at baseline) instead of ΔHBF, in order to control for baseline HBF. A multiple regression was run to predict ΔHBF from sex, APOE4 carrier status, ΔSBP, and the interaction between ΔSBP and APOE4 carrier status. Age, ΔDBP and the interaction term for ΔDBP and APOE4 carrier status worsened the predictive value and were therefore excluded from the model. Two-way ANCOVAs were utilized to examine the effects of APOE4 carrier status and Treatment Group (exercise or control) on change in cognitive functioning (visuospatial, executive and memory), ΔSBP, ΔDBP, change in hippocampal volume, change in body mass index (BMI), and change in VO2max (ΔVO2mx), after controlling for age and sex. A Pearson’s Product-Moment Correlation assessed the relationship between ΔHBF and change in verbal memory function (ΔVM) for the APOE4 carriers who underwent the exercise intervention.

Data availability policy

This trial was prospectively registered with ClinicalTrials.gov (NCT02000583). Anonymized data will be shared by request from any qualified investigator.

Results

A total of 109 participants (93%: control n = 34, aerobic exercise n = 75) completed the study (Figure 1). Genotyping was not completed for 3 participants. APOE2 carriers (n = 14) and participants with incomplete ASL MRI data (n = 4) were excluded from the present analysis. Of the remaining 88 participants, 44 had hypertension at the baseline visit, classified due to having SBP ≥130 mmHg (n = 22), DBP ≥80 mmHg (n = 5), or both (n = 17).²⁶ Sixteen participants completed less than 80% of the prescribed exercise, but all participants were included in the analyses regardless of the percent exercise intervention prescription completed.

Figure 1.

Flowchart of Alzheimer’s Prevention through Exercise (APEx) study participants.

Baseline characteristics of hypertensive and normotensive participants

At baseline, HBF was significantly lower in the hypertensive participants (30.17 ± 6.06 mL/100g/min) than the normotensive participants (34.00 ± 9.54 mL/100g/min) (p = 0.028). A multiple linear regression showed this difference was driven by baseline SBP. Specifically, the regression analysis, controlling for age, sex and APOE4 carrier status, significantly predicted HBF (F(5,82) = 2.657, p = 0.028) with SBP contributing significantly to the model (B = −0.157, p = 0.035) and DBP having a non-significant effect (B = 0.086, p = 0.461). There were no significant differences between the hypertensive and normotensive participants in age, sex, education, VO₂max or hippocampal volume (see Table 1).

Table 1.

Baseline characteristics by group.

	A) Non-carrier, Control (n = 7)	B) Non-carrier, Exercise (n = 15)	C) APOE4 Carrier, Control (n = 8)	D) APOE4 Carrier, Exercise (n = 14)	p-value Groups A-D	E) HTN Total (A + B+C + D) (n = 44)	F) NTN (n = 44)	p-value Groups E&F
Age, years	72.4 (5.1)	70.9 (4.5)	72.9 (6.8)	71.9 (6.0)	0.895	71.8 (5.4)	71.2 (4.8)	0.602
Males, n (%)	3 (43%)	7 (47%)	2 (25%)	5 (36%)	0.768	17 (39%)	12 (27%)	0.257
Education, years	16.0 (2.3)	16.0 (2.8)	16.5 (1.7)	16.1 (1.4)	0.851	16.1 (2.1)	16.0 (2.4)	0.706
Anti-HTN med, n (%)	4 (57%)	6 (40%)	3 (38%)	7 (50%)	0.861	20 (46%)	16 (36%)	0.516
APOE4 carriers, n (%)	0 (0%)	0 (0%)	8 (100%)	14 (100%)	−	22 (50%)	22 (50%)	1.000
Baseline SBP, mmHg	144.1 (13.5)	138.4 (10.3)	133.5 (5.4)	141.0 (14.0)	0.254	139.2 (11.6)	118.1 (8.5)	<0.001*
Baseline DBP, mmHg	79.9 (8.4)	81.0 (9.5)	78.0 (7.7)	81.1 (8.6)	0.920	80.3 (8.5)	70.0 (5.9)	<0.001*
Baseline HBF, mL/100g/min	30.40 (4.94)	30.96 (6.36)	33.02 (5.67)	27.59 (6.04)	0.207	30.17 (6.06)	34.00 (9.54)	0.028*
Baseline VO2max, mL/kg/min	19.76 (3.34)	23.33 (6.56)	22.58 (4.08)	21.56 (4.65)	0.482	22.06 (5.13)	23.15 (5.20)	0.325
Baseline HV, mL	7.43 (0.94)	7.34 (0.89)	7.19 (1.17)	7.15 (0.87)	0.906	7.27 (0.92)	7.57 (0.78)	0.096

^*Significant (p < 0.05).

Note: Values are shown as mean (standard deviation) unless otherwise indicated. HTN Total (E) combines groups A-D. NTN (F) is a separate group. Statistical tests were applied to compare baseline characteristics among groups A-D (p-value Groups A-D) or between groups E&F (p-value Groups E&F).

NTN: normotensive (participants with normal blood pressure at baseline); HTN: hypertensive (participants with hypertensive blood pressure at baseline); Anti-HTN med: taking anti-hypertension medication; SBP: systolic blood pressure; DBP: diastolic blood pressure; HBF: hippocampal blood flow; VO₂max: maximal oxygen uptake; HV: hippocampal volume.

Baseline characteristics by intervention and APOE4 carrier status

Only the participants with hypertension at the baseline visit (n = 44) were included in further analyses. Half of these participants were APOE4 carriers (APOE3/APOE4, n = 20; APOE4/APOE4, n = 2). At baseline, there were no significant differences among APOE4 non-carriers assigned to the control group (n = 7), APOE4 non-carriers assigned to exercise (n = 15), APOE4 carriers assigned to the control group (n = 8), and APOE4 carriers assigned to exercise (n = 14) in age, sex, education, anti-hypertensive medication use, SBP, DBP, HBF, maximal oxygen uptake (VO₂max), or hippocampal volume (Table 1).

ΔHBF with exercise intervention

The two-way ANCOVA met the assumptions of homoscedasticity and homogeneity. There were no outliers in the data, as assessed by no cases with studentized residuals greater than ±3 SD. Studentized residuals were normally distributed, as assessed by Shapiro-Wilk's test (p > .05). Means, adjusted means (for age and sex), SD and standard errors are presented in Table 2 and shown graphically in Figure 2. There was a significant two-way interaction between APOE4 carrier status and Treatment Group on ΔHBF, while controlling for age and sex, F(1, 38) = 4.504, p = 0.040, partial η² = 0.106. Therefore, an analysis of simple main effects for APOE4 carrier status and Treatment Group was performed with statistical significance receiving a Bonferroni adjustment and being accepted at the p < 0.025 level.

Table 2.

Mean change in HBF (ΔHBF) from baseline to 52 weeks.

ΔHBF (%ΔHBF) mL/100g/min (%)	Non-carrier Control (n = 7)	Non-carrier Exercise (n = 15)	APOE4 Carrier Control (n = 8)	APOE4 Carrier Exercise (n = 14)
Mean	−0.19 (−0.2%)	−0.48 (−1.8%)	−2.22 (−5.5%)	+4.05 (+17.1%)
(SD)	2.14	4.06	4.32	6.21
Meanadjusted	−0.32 (−0.8%)	−0.54* (−2.1%)¥	−2.08+ (−4.8%)‡	+4.09*,+ (+17.3%)¥,‡
(SE)	1.79	1.23	1.69	1.26

SD: standard deviation; SE: standard error; ^*p: 0.013; ⁺p: 0.006; ^¥p: 0.005; ^‡p: 0.007.

Note: The mean change value (ΔHBF in mL/100g/min) is followed by mean percent ΔHBF (%ΔHBF) for each group. Mean_adjusted represents the group means, adjusted for age and sex, for participants divided by APOE4 carrier status and clinical trial arm (exercise intervention or control). Significant results of the simple main effects analysis that followed up the significant two-way ANCOVA interaction are indicated by superscripts (^*,+,¥,‡). All other pairwise comparisons were not significant.

Figure 2.

Mean change in hippocampal blood flow (ΔHBF) from baseline to 52 weeks. (a) Among APOE4 carriers, those who underwent the exercise intervention had a significantly larger ΔHBF (increased HBF) over the 52 weeks than the control group (p = 0.006). Additionally, within the exercise intervention arm, ΔHBF was significantly larger for the APOE4 carriers than the non-carriers (p = 0.013). There were no other significant differences between groups.

X = mean; horizontal line = median

(b) Representative arterial spin labeling MRI (ASL-MRI) scans from a participant with high (top panel) and a participant with low (bottom panel) HBF.

The effect of Treatment Group on ΔHBF was significant for the APOE4 carriers, F(1, 38) = 8.597, p = 0.006, partial η² = 0.184. Specifically, for the APOE4 carriers, adjusted mean ΔHBF was higher for participants who underwent the exercise intervention (4.09 mL/100g/min) than for the control group (−2.08 mL/100g/min), a significant difference of 6.17 (95% CI, 1.91 to 10.44) mL/100g/min. The effect of Treatment Group was not significant in the APOE4 non-carriers, F(1, 38) = 0.011, p = 0.918, partial η² < 0.0001.

The effect of APOE4 carrier status on ΔHBF in the exercise group was significant, F(1, 38) = 6.853, p = 0.013, partial η² = 0.153. Specifically, for participants who underwent the exercise intervention, adjusted mean ΔHBF was higher for the APOE4 carriers (4.09 mL/100g/min) than the non-carriers (−0.54 mL/100g/min), a significant difference of 4.63 (95% CI, 1.05 to 8.22) mL/100g/min. In the control group, the effect of APOE4 carrier status on ΔHBF was not significant, F(1, 38) = 0.514, p = 0.478, partial η² = 0.013.

Overall, the 52-week exercise intervention improved mean HBF for the APOE4 carriers from 27.59 to 31.64 mL/100g/min. Over this same time period, the APOE4 carriers in the control group experienced a decline in HBF from 33.02 to 30.80 mL/100g/min, and the APOE4 non-carriers remained relatively stable in both the control (30.40 to 30.22 mL/100g/min) and exercise intervention (30.96 to 30.48 mL/100g/min) groups. It is thus noteworthy that the APOE4 carriers in the exercise intervention had the lowest mean HBF at baseline but the highest HBF after the intervention, although these groups differences were not significant at baseline (p = 0.207) or post-intervention (p = 0.961).

%ΔHBF with exercise intervention

There was a significant two-way interaction between APOE4 carrier status and Treatment Group on %ΔHBF, while controlling for age and sex, F(1, 38) = 4.392, p = 0.043, partial η² = 0.104. The effect of Treatment Group on %ΔHBF was significant for the APOE4 carriers, F(1, 38) = 8.002, p = 0.007, partial η² = 0.174. Specifically, for the APOE4 carriers, adjusted mean %ΔHBF was higher for participants who underwent the exercise intervention (+17.3%) than for the control group (−4.8%), a significant difference of 22.1% (95% CI, 6.3% to 38.0%). The effect of Treatment Group was not significant in the APOE4 non-carriers, F(1, 38) = 0.028, p = 0.869, partial η² = 0.001. The effect of APOE4 carrier status on %ΔHBF in the exercise group was significant, F(1, 38) = 8.723, p = 0.005, partial η² = 0.187. Specifically, for participants who underwent the exercise intervention, adjusted mean %ΔHBF was higher for the APOE4 carriers (+17.3%) than the non-carriers (−2.1%), a significant difference of 19.4% (95% CI, 6.1% to 32.7%). In the control group, the effect of APOE4 carrier status on %ΔHBF was not significant, F(1, 38) = 0.197, p = 0.660, partial η² = 0.005.

ΔHBF with change in systolic blood pressure (ΔSBP)

A multiple regression was run to predict ΔHBF from sex, APOE4 carrier status, ΔSBP, and the interaction between ΔSBP and APOE4 carrier status. The regression analysis met the assumptions of linearity, normality, homoscedasticity, and independence of residuals. There was no multicollinearity. There were no outliers, as assessed by no studentized deleted residuals greater than ±3 SD. The multiple linear regression model significantly predicted ΔHBF, F(4, 39) = 3.134, p = 0.025, R² = 0.243, adjusted R² = 0.166. Regression coefficients and standard errors can be found in Table 3.

Table 3.

Regression model for 52-week change in hippocampal blood flow (ΔHBF).

Variable	B	SEB	β	p-value
Intercept	−1.529	1.232		0.222
Sex (Male)	2.369	1.478	0.226	0.117
ΔSBP	−0.015	0.065	−0.041	0.815
APOE4 carrier status (+)	2.479	1.458	0.243	0.097
ΔSBP x APOE4 carrier status (+)	−0.243	0.111	−0.376	0.035*

^*Significant (p < 0.05), ΔSBP = change in systolic blood pressure.

Note: The multiple linear regression model for participants with baseline hypertension (N = 44) significantly predicted ΔHBF (mL/100 g/min), F(4, 39) = 3.134, p = 0.025, R² = 0.243, adjusted R² = 0.166.

There was a significant interaction between ΔSBP and APOE4 carrier status on ΔHBF (p = 0.035). The relationship between ΔSBP and ΔHBF is shown graphically for APOE4 carriers and non-carriers in Figure 3. For APOE4 carriers, reductions in SBP over the year of the study resulted in higher HBF, while increases in SBP (reflecting further elevation of already elevated blood pressure) resulted in decreased HBF. In contrast, there was no relationship between changes in SBP and HBF for the APOE4 non-carriers.

Figure 3.

Relationship between change in systolic blood pressure (ΔSBP) and hippocampal blood flow (ΔHBF). Over the 52-week clinical trial period, ΔSBP was inversely associated with ΔHBF for the APOE4 carriers. There was no relationship between ΔSBP and ΔHBF for the APOE4 non-carriers. This suggests reductions in SBP in older adults with baseline hypertension may preferentially improve HBF for APOE4 carriers.

Cognitive function changes

There was no significant two-way interaction between APOE4 carrier status and Treatment Group on change in visuospatial functioning (p = 0.755), executive functioning (p = 0.841), or verbal memory (ΔVM, p = 0.434), suggesting no group differences in mean change in cognitive scores from baseline to post-intervention. However, there was a significant positive correlation between ΔHBF and ΔVM for the APOE4 carriers who underwent the exercise intervention (r = 0.561, p = 0.037). That is, improvements in HBF from the exercise intervention correlated with improved verbal memory performance in APOE4 carriers.

Other physiological changes

There was no significant two-way interaction between APOE4 carrier status and Treatment Group for ΔSBP (p = 0.058), ΔDBP (p = 0.260), or change in hippocampal volume (p = 0.767). There was no significant two-way interaction between APOE4 carrier status and Treatment Group on change in BMI (p = 0.921). However, the main effect of Treatment Group on change in BMI was significant (p = 0.015). Specifically, participants who exercised reduced their BMI on average over the year (mean change of –0.540 kg/m²) while participants in the control group gained weight (mean change of +0.185 kg/m²). Likewise, there was no significant two-way interaction between APOE4 carrier status and Treatment Group on ΔVO₂max (p = 0.116) but the main effect of Treatment Group on ΔVO₂max was significant (p = 0.002). Participants who underwent the exercise intervention had a mean ΔVO₂max of 2.73 mL/kg/min, which was significantly higher than the control group mean ΔVO₂max of 0.41 mL/kg/min. These findings show the exercise intervention improved cardiorespiratory fitness and reduced BMI and that these effects were not different between APOE4 carriers and non-carriers.

Discussion

In this analysis of a secondary outcome from a randomized controlled trial, we report that an aerobic exercise intervention selectively improved hippocampal blood flow (HBF) for hypertensive APOE4 carriers. Additionally, we demonstrate that reductions in systolic blood pressure (SBP) for hypertensive individuals were tied to improvements in HBF for APOE4 carriers only. Finally, we found that these improvements in HBF were correlated with improved verbal memory performance.

People with Alzheimer’s disease (AD) have lower cerebral blood flow (CBF) than age-matched controls.^2,30 A recent study involving over 7,700 scans from 1,171 people in the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database found cerebrovascular dysregulation was the earliest pathological event during AD development, followed by changes in β-amyloid deposition, metabolic dysfunction, functional impairment and structural atrophy.¹ Therefore, interventions that maintain or improve CBF with aging may prevent or delay dementia development, and this may be particularly true for regions closely involved in the disease process, such as the hippocampus.^31,32 Indeed, one recent study of almost 250 participants found APOE4 carriers exhibited increased blood brain barrier breakdown specifically in the hippocampus and medial temporal lobe.⁴ This hippocampal cerebrovascular dysfunction was present even in APOE4 carriers who had normal cognition, manifested independently of β-amyloid and tau pathology, and showed increasing severity with cognitive decline.⁴ Additionally, higher blood brain barrier breakdown at baseline predicted future cognitive decline for the APOE4 carriers only.⁴ This blood brain barrier breakdown has been associated with reduced CBF,³³ suggesting impaired hippocampal blood flow (HBF) could be an important mechanism early in the pathogenesis of AD for APOE4 carriers. This mechanism may therefore provide a novel therapeutic target to prevent or treat AD for those at highest genetic risk.

In the present study, we employed aerobic exercise as a therapeutic approach to prevent AD, and we report that aerobic exercise improved blood delivery to the hippocampus (HBF) for hypertensive APOE4 carriers. Specifically, we found a mean HBF increase (+17.3%) for APOE4 carriers who exercised compared to a mean HBF decline (−4.8%) for APOE4 carriers in the control group over one year. This HBF improvement could prevent or delay AD for these individuals at highest known genetic risk, considering the growing evidence that CBF reductions precede measurable cognitive decline^1,34 and likely contribute causally to dementia pathogenesis,^3,4,33,35,36 particularly for APOE4 carriers.^13,35,37 Indeed, APOE4 carriers have been shown to experience an accelerated age-related CBF decline,^38,39 and some studies suggest cerebrovascular dysfunction may act synergistically with the APOE4 allele to promote cognitive decline.^4,13,37 If true, this would mean CBF maintenance is even more important for APOE4 carriers than non-carriers in order to prevent dementia, strengthening the clinical relevance of our current findings.

An observational study published earlier this year combined data from the Chicago Health and Aging Project and the Memory and Aging Project and concluded that adults should participate in at least 150 minutes of moderate-to-vigorous physical activity per week to reduce dementia risk,⁴⁰ which is notably the same exercise prescription utilized in the intervention arm of our trial. Previous randomized controlled trials have shown promising results for exercise in preventing cognitive decline,^9,19,24,41 and a recent systematic review concluded exercise improves cognitive function in people 50+ years of age, independent of baseline cognitive status.⁴² A better understanding of the mechanisms through which this occurs would allow tailoring and targeting of exercise interventions for populations most likely to benefit. One important mechanism may be through improvements in CBF, particularly in the hippocampus. Indeed, aerobic exercise has been shown to increase HBF both acutely ⁴³ and chronically,^8,10,44 with increased HBF correlating with cognitive gains.^11,12 Here, we add to the literature by reporting a selective increase in HBF with chronic exercise for APOE4 carriers. The positive correlation between ΔHBF and ΔVM for our APOE4 intervention group suggests these improvements in HBF may play a role in preventing cognitive decline.

There have been conflicting reports on the influence of the APOE4 allele in exercise-induced changes in cognitive function and brain health. One randomized trial involving 170 older adults with self-reported memory problems suggested less cognitive benefit from exercise for APOE4 carriers.²⁴ However, a more recent trial of 200 patients with mild AD found APOE4 carriers experienced more improvement in cognitive function after an exercise intervention than non-carriers.¹⁹ Observational studies have likewise produced conflicting findings. For example, one study of cognitively normal older adults concluded high physical activity level may preserve hippocampal volume in APOE4 carriers selectively.²¹ However, a different cross-sectional study including older adults with and without AD found APOE4 carrier status did not influence the relationship between cardiorespiratory fitness and brain atrophy.⁴⁵ Our current findings expand upon these conflicting results by providing further evidence for a preferential benefit of exercise for APOE4 carriers, specifically in HBF improvement. We likely observed this preferential benefit for APOE4 carriers in the current study due to our selected outcome measure (HBF). There is increasing evidence that AD manifests differently for APOE4 carriers and non-carriers, with APOE4 carriers experiencing early pathology in the hippocampus associated with verbal memory impairment, while non-carriers with AD tend to have early pathology in the frontal and parietal lobes associated with impaired visuospatial and executive functioning.⁴⁶ Additionally, APOE4 carriers with normal cognition have greater blood brain barrier breakdown in the hippocampus than non-carriers, and this cerebrovascular dysfunction is more predictive of future cognitive decline in APOE4 carriers than non-carriers.⁴ This suggests that distinct mechanisms may be involved in the pathogenesis of AD depending on APOE4 genotype. Thus, we may have detected a benefit in HBF from the exercise intervention specifically for APOE4 carriers because cerebrovascular dysfunction in this region plays a more important role for participants with this genotype, who are therefore primed to show the greatest benefit. Additionally, hypertension has been repeatedly shown to be more detrimental for APOE4 carriers than non-carriers,^16,17 which means the hypertensive APOE4 carriers in our study were likely at greater risk of brain pathology than the hypertensive APOE4 non-carriers.

Hypertension may act mechanistically through cerebrovascular dysfunction to cause brain pathology.⁴⁷ Unlike family history, hypertension is a modifiable risk factor for dementia, and lowering blood pressure is thus an enticing intervention strategy to slow or prevent cognitive decline. The landmark Systolic Blood Pressure Intervention Trial Memory and Cognition in Decreased Hypertension (SPRINT-MIND) trial published in 2019 demonstrated that aggressive BP lowering (SBP < 120 mmHg) through medication significantly reduced the risk of mild cognitive impairment (MCI) in 9,361 older adults.⁴⁸ Notably, this was the first instance of an intervention of any type effectively reducing MCI incidence in a large population.⁴⁸ Although the mechanism through which aggressive SBP lowering prevented cognitive decline was beyond the scope of the SPRINT-MIND trial, other studies have suggested reducing BP in people with hypertension may act mechanistically by augmenting CBF. For example, one group found intensive BP lowering in cognitively-normal older adults significantly increased gray matter CBF.⁴⁹ More recently, the Nilvadipine in AD (NILVAD) trial reported that reducing SBP significantly increased HBF in adults with mild-to-moderate AD, suggesting SBP reductions may provide benefit by increasing blood flow to the hippocampus specifically.⁵⁰ In the current study, baseline assessment of the entire population (hypertensive and normotensive participants, n = 88) showed higher SBP was related to lower HBF, even when controlling for age, sex and APOE4 carrier status. This finding provides further evidence for the connection between hypertension and reduced HBF in older adults.

Notably, it is unclear whether lowering BP provides the same benefit regardless of the intervention (i.e. medication, exercise, dietary changes). The SPRINT-MIND trial allowed a variety of medications to reach target SBP, including thiazide-type diuretics, calcium channel blockers, angiotensin converting enzyme inhibitors, angiotensin receptors blockers and more.⁴⁸ Therefore, the observed effect of decreased MCI incidence was unlikely to have been tied to a specific mechanism of a single medication and more likely to have resulted directly from reduced SBP. Additionally, in the present study we found a significant relationship between reductions in SBP and improved HBF with aerobic exercise, while the NILVAD trial found reductions in SBP associated with improved HBF with a specific medication, nilvadipine.⁵⁰ We thus propose that the common mechanism may be reduced SBP, which results in cerebrovascular remodeling and improved blood flow delivery. Still, it is likely that both exercise and medications have unique effects throughout the body beyond BP reduction, and further exploration is needed to identify the ideal method for lowering BP in order to improve brain health.

By selecting only individuals with hypertension at baseline (defined using the ACC/AHA Clinical Practice Guidelines²⁶) the present analysis included older adults at high baseline vascular risk. We found that reductions in SBP for these participants with initially elevated BP were tied to improvements in HBF for the APOE4 carriers only. This is in line with previous literature showing elevated blood pressure acts synergistically with APOE4 carrier status to impair cognitive function¹⁷ and promote brain pathology.¹⁶ Furthermore, while the NILVAD sub-study did not assess APOE4 carrier status,⁵⁰ it seems plausible that the observed improvement in HBF from SBP reduction could have been driven by a large proportion of APOE4 carriers in the study population, considering the high prevalence of APOE4 in AD (∼65% of people with AD compared to ∼25% of the general population).⁵¹ Regardless, the existing literature combined with our current data provide strong evidence for a synergistic relationship between the APOE4 allele and peripheral BP on promoting brain pathology and cognitive dysfunction, suggesting interventions to lower BP are particularly important for this patient population.

It is important to note that although we observed a significant improvement in HBF only for APOE4 carriers with baseline hypertension, this does not imply that exercise lacks beneficial value for APOE4 carriers without hypertension. Clinical trials such as the present study cover a relatively short, defined time period during which it can be challenging to elicit significant measurable change. For this reason, clinical trials often select participants at elevated risk of illness to increase the likelihood of observing an effect of the intervention. Thus, in the present analysis we selected APOE4 carriers with baseline hypertension, which places them at elevated vascular risk, because we sought to assess a vascular outcome, HBF. Therefore, it is likely that we observed significant differences with the intervention specifically in the hypertensive group because this was a particularly at-risk population for cerebrovascular dysfunction. That is, these participants with poor vascular health were primed to have the greatest (and therefore, most obvious) cerebrovascular benefits from aerobic exercise, as this intervention is known to induce profound, beneficial vascular adaptations.⁵ Additionally, it is possible that APOE4 carriers with normotension at baseline experienced significant improvements in other areas that were not captured through an analysis of HBF. For example, aerobic exercise has been shown to reduce hippocampal inflammation in rats,^52,53 an effect that would not have been captured through ASL-MRI utilized in the current study. Therefore, future studies are needed to assess changes in domains beyond brain blood flow that may contribute to cognitive improvements observed with aerobic exercise.

Our study has a number of limitations. This was an analysis of a secondary outcome of a clinical trial with different primary aims,²⁵ and the findings should therefore be interpreted with caution. All participants were healthy and cognitively normal, which may complicate the generalizability of our data to patient populations. Although changes in verbal memory did correlate with changes in HBF over the 52-week period for the APOE4 carriers who exercised, there were no differences among intervention groups in mean change in cognitive functioning for the three domains, which means we cannot state that exercise improved cognition. The ACC/AHA guidelines recommend BP measurements taken on at least two separate occasions in order to diagnose hypertension. For the current study, we obtained two BP measurements on a single occasion. Therefore, some participants may have been included in our hypertensive group who would not have met the criteria if given a second reading on a separate occasion. We analyzed data from all participants regardless of the amount of exercise completed, which does not account for the potential effect of real exercise dose on HBF (only prescribed dose). In future studies, implementation of wearable activity monitors in both the intervention and control arms could allow for better characterization of this relationship. As we did not perform a lipid panel or arterial stiffness assessment on these participants, we could not assess the impact of the aerobic exercise intervention on changes in arterial stiffness or cholesterol levels, both of which likely played a role in mediating some of the observed exercise effects. Finally, although our trial included a longer intervention than the majority of previously-published exercise intervention trials,^{6,9,11,19,24,41} the follow-up period of 52 weeks may still be too short to sufficiently characterize long-term clinical relevance of the exercise intervention. Future studies that follow participants for many years post-intervention would be better suited to assess clinical implications such as dementia risk.

In the current study, we report an aerobic exercise intervention improved HBF for cognitively normal hypertensive APOE4 carriers (but not non-carriers) and that greater HBF improvement over the 52-week intervention period related to better verbal memory performance for this group. Additionally, we show reductions in SBP may improve HBF for APOE4 carriers only. These findings suggest aerobic exercise interventions, especially those that lower SBP, may be particularly beneficial for APOE4 carriers with baseline hypertension. This knowledge could inform the design and execution of future interventional trials.