Comparing the Effects of Two Cardiovascular Health Factors on Working Memory Capacity in Healthy Aging: Separate and Combined Effects of Arterial Elasticity and Physical Fitness

Abstract

Objectives

Arterial elasticity and physical fitness are 2 important cardiovascular health factors that influence cognition in older adults. Working memory capacity (WMC), a core component underlying cognitive aging across many cognitive domains, may be affected by individual differences in cardiovascular health in older adults. This study aims to identify in older adults: (a) separate and combined effects of these 2 cardiovascular health factors on WMC and (b) which of the 2 factors is more critical in influencing WMC.

Methods

WMC in 89 healthy older adults was assessed by 2 complex span tasks. Arterial elasticity was assessed by pulse pressure (PsP). Physical fitness was measured by an established proxy of VO₂ max (MET). Effects of PsP and MET on WMC were evaluated via step-wise regressions.

Results

After controlling for age, sex, and education, PsP and MET were separately predictive of WMC in older adults. Together, the combined effect of PsP and MET was more predictive of WMC than fitness alone, but not more than PsP alone. Mediation analysis indicates that the relationship between MET and WMC was completely mediated by PsP.

Discussion

This study innovatively demonstrates that though arterial elasticity and physical fitness separately predict WMC, the former completely mediates the relationship between fitness and WMC. This suggests that biologically based cardiovascular health factors like arterial elasticity are crucial individual difference variables that should be measured and monitored in cognitive aging studies as well as in physical interventions that are designed to improve cognition in healthy aging.

Many daily activities require us to keep track of or to remember multiple sources of information, such as watching a movie with nonlinear plots, driving in a metropolitan on a busy multilane highway, and cooking complex recipes that need multiple ingredients to be added at different times. The largest number of items one can actively maintain, coordinate, and retrieve is considered as the capacity of working memory for an individual. For most adults, working memory capacity (WMC) is quite limited and ranges from one to four items depending on sequential or simultaneous processing of information (Basak & Verhaeghen, 2003, 2011a, 2011b; Cowan, 2001). As with other basic cognitive abilities, WMC declines rapidly with age (for meta-analysis and review see Bopp & Verhaeghen, 2005; Borella et al., 2008). However, there is a tremendous amount of variance regarding WMC in older adults (aged 60 and older). In one study, the average capacity of old (2.07) was almost a unit smaller than the young (2.83), with some older adults having the same capacity as younger adults and also exhibiting similar retrieval speed of information within the focal capacity (Basak & Verhaeghen, 2003). These results suggest that some older adults not only have the same capacity as younger adults, but they also perform with equivalent efficiency. It is of great importance to identify the protective factors that may help us maintain our WMC with the same efficiency well into our late adulthood.

A recent review on working memory performance examined 21 factors that may affect WMC in the healthy life-span population and showed that better cardiovascular health was consistently associated with higher WMC (Blasiman et al., 2018). Therefore, we hypothesize that cardiovascular health factors may mitigate the detrimental effects of age on WMC in healthy aging. In this study, we therefore conducted an in-depth investigation focusing on not one, but two cardiovascular health factors—physical fitness and arterial wall elasticity—that have been touted for their clinical implications, and their relationships with WMC in healthy aging.

Older adults tend to become increasingly sedentary as they age (Diaz et al., 2016), which has implications on their cardiovascular health. Low physical fitness, indicated by low VO₂ max (or VO₂ peak), is considered to be a key cardiovascular risk in older adults. Studies have reported an advantage of fitness on WMC in older adults, with high-fit older adults (i.e., regular exercisers) showing higher WMC compared to low-fit older adults (Chang et al., 2013; Newson & Kemps, 2008; Shin et al., 2012). In addition to these cross-sectional studies that investigated fitness as a between-group factor, physical exercise training studies with healthy older adults have consistently demonstrated that increased exercise improves older adults’ cognitive performance, including WMC (for a recent meta-analysis see Falck et al., 2019).

On top of reduced fitness with age, epidemiological studies have consistently reported that older adults have proportionally less arterial elasticity (indicated by increased wall thickening), which puts them at greater risks to chronic obstructive pulmonary disease (Vivodtzev et al., 2014), heart diseases (Reneman et al., 2006), and other cardiovascular diseases (Cecelja & Chowienczyk, 2012). Decreases in arterial wall elasticity lead to increases in blood pressure and pulse pressure (PsP; Safar et al., 2012). High blood pressure has been associated with poorer cognitive performance in middle-aged and older adults (Gayda et al., 2017; Ihle et al., 2017; Takeda et al., 2017). Furthermore, longitudinal studies in older adults have shown that high blood pressure predicts significant declines in WMC over 5 years (Elias et al., 2010; Raz et al., 2007). All previous studies on WMC have used blood pressure as an indirect marker of arterial elasticity. However, the American Heart Association (Whelton et al., 2018) suggests that PsP (systolic − diastolic) is a better measure of arterial elasticity than systolic or diastolic blood pressure alone.

Although the effects of cardiovascular health factors, such as fitness and arterial elasticity, on WMC in older adults seem well studied, there has yet to be a study investigating both health factors—comparing which one is more important in influencing WMC, as well as finding out the combined effects of multiple cardiovascular health factors on WMC in older adults. All past studies have investigated the effects of just one health factor (fitness or elasticity) on WMC in older adults (Chang et al., 2013; Elias et al., 2010; Gayda et al., 2017; Ihle et al., 2017; Newson & Kemps, 2008; Raz et al., 2007; Shin et al., 2012). The potential mediating effects of one health factor on the relationship between other health factors and cognition were scarcely investigated. Epidemiological studies have suggested a moderate correlation between fitness and elasticity measures in older adults (Deiseroth et al., 2019). It is possible that the relationship between fitness and WMC could be mediated by arterial elasticity, similar to neuronal metabolite concentration mediating the relationship between fitness and WMC (Erickson et al., 2012). A mediation effect would suggest an indirect benefit of fitness on WMC through benefitting arterial elasticity.

Furthermore, investigations into which cardiovascular health factor contributes more in affecting WMC in older adults could help clinicians and health care facilities develop targeted intervention strategies in improving cognitive performance in older adults. Essentially, should older adults aim to improve working memory focus on maintaining healthy blood pressure or improving cardiac output through exercise or both?

It is important to note that all aforementioned studies, investigating effects of fitness or blood pressure on WMC in older adults, used either simple digit span tasks or visual short-term memory tasks to assess WMC. Working memory was defined by Baddeley as the relationship between “passive storage and active manipulation or transformation of information” (Baddeley, 1992). The limited capacity of working memory for any individual could be most accurately measured by tasks that require one to actively manipulate and transform information to and from passive storage. Although simple digit span tasks require “passive storage of information,” which is a hallmark of short-term memory, they do not accurately capture the transformation and manipulation aspect of working memory. Complex span tasks (e.g., computation span and reading span) are dual-task paradigms, where one has to process and make comparisons about some information, followed by remembering different information. Such dual-task working memory paradigms therefore measure individual WMC more accurately than simple digit span tasks. In older adults, the degree of impairments in WMC, measured by complex span tasks, would affect the outcome of any study investigating age-related differences using complex tasks with increasing working memory load (for reviews see Basak & Zelinski, 2013; Stine, 1995).

This study is the first to examine the effects of multiple cardiovascular health factors on WMC, measured by complex span tasks, in older adults. Past researches have investigated the effects of arterial elasticity or fitness on WMC separately of each other. Because both arterial elasticity and physical fitness are measures of cardiovascular health, they are likely correlated with each other. Therefore, it is critical to investigate the combined effects of both factors on WMC and to examine if the two cardiovascular health factors have an additive effect on WMC in older adults. We hypothesize that both arterial elasticity and physical fitness would be separately associated with WMC in older adults. However, the combination of high arterial elasticity and high physical fitness would have a higher predictive power of larger WMC, compared to the predictive power of either factor alone. Although given the potential intercorrelation between these two health factors, it is possible that one of them is driving the combined effect. If the two factors are indeed correlated, one factor may mediate the relationship between the other and WMC. Therefore, the predictive power of arterial elasticity and physical fitness on WMC in older adults will be compared. Such a unique approach would not only further our understanding of individual differences in WMC, but can also have important clinical implications regarding the selection of the relevant cardiovascular health factor in order to combat age-related cognitive declines.

Method

Participant Characteristics

Eighty-nine participants older than the age of 60 years were recruited from the Dallas–Fort Worth metroplex through flyers and newspaper advertisements. Participants were screened for any medical, neurological, or psychiatric illnesses. Other exclusion criteria included current or previous substance abuse, depression (assessed by the Geriatric Depression Scale, Burke et al., 1991), psychiatric disorders, and a Montreal Cognitive Assessment (MoCA) score of less than 24. A MoCA cutoff score of 24 was determined based on studies in older adults with cardiovascular conditions (McLennan et al., 2011; Potocnik et al., 2020). These studies across multiple countries have shown that a MoCA score below 24 (instead of 26) should be used to categorize older adults with existing cardiovascular conditions as having mild cognitive impairment.

All participants were native or fluent English speakers, had completed high school or 12 years of formal education, and had normal or corrected 20/30 vision. Thirty-eight participants reported taking medications or supplements for blood pressure. This study was approved by the University of Texas at Dallas Institutional Review Board.

Analysis of Statistical Power

We planned on conducting a series of regression analyses to investigate: (a) the separate effects of arterial elasticity or physical fitness on WMC in older adults (R² deviation from zero) and then (b) the combinatorial effect of arterial elasticity and physical fitness on WMC over and beyond the effects observed in the first set of analyses, where the separate effects of these factors on WMC were assessed (R² increase with three predictors). Power analyses were therefore computed for linear multiple regressions; first for R² deviation from zero and second for R² increase.

We selected an effect size for power calculations based on a study in younger adults using an operation span task to assess WMC (Padilla et al., 2014). In the study of Padilla et al., high-fit younger adults had a larger operation span compared to low-fit younger adults, with a Cohen’s d of 0.56 that corresponded to an f² of 0.09. We selected this study on younger adults because most studies assessing effects of cardiovascular health factors on WMC in older adults used only backward digit span, which is simpler than the complex span tasks used in this study, based on a meta-analysis where complex spans show greater age-related deficits than backward span (Bopp & Verhaeghen, 2005). Power analyses were conducted using G*Power 3.1 package (Faul et al., 2007), with a Type I error rate of 0.05.

For regression analysis with one predictor (R² deviation from zero), a sample of N = 89 guarantees a power of 80% to find a significant R² deviation from zero of one predictor for the effect size of f² = 0.09. For regression analysis with three predictors (R² increase), a sample of N = 89 has 63% power to detect a significant R² increase for the same effect size.

We expect the effects of cardiovascular health factors on WMC to be larger in older adults than in a group of healthy undergraduate students (as assessed by Padilla et al., 2014). Therefore, our chosen effect size, d = 0.56 (f² = 0.09), should be considered to be a very conservative estimate.

Assessments of Arterial Elasticity and Physical Fitness

Blood pressure, height, weight, and resting heart rate were taken from every older participant. Arterial elasticity was measured via PsP. PsP has been suggested as an important predictor of future cardiovascular diseases in healthy older adults (Alderman, 1999; Blacher et al., 2000). For older hypertensive patients taking medication, PsP is the single strongest predictor of future cardiovascular complications (Blacher et al., 2000). PsP for each participant was calculated through the following equation:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{P}\mathrm{s}\mathrm{P}=\mathrm{s}\mathrm{y}\mathrm{s}\mathrm{t}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{b}\mathrm{l}\mathrm{o}\mathrm{o}\mathrm{d}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}-\mathrm{d}\mathrm{i}\mathrm{a}\mathrm{s}\mathrm{t}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{b}\mathrm{l}\mathrm{o}\mathrm{o}\mathrm{d}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}}\end{array}}$$

High PsP indicates lower arterial elasticity.

An independent sample t-test between older adults who reported taking medication or supplements (N = 38) and older adults who did not report taking any medication or supplement for blood pressure (N = 51) showed a significant difference in PsP (t = 2.74, p < .01), but not in systolic (t = 1.25, p = .21) or diastolic (t = −1.89, p = .06) blood pressures. Older adults taking medications (or supplements) had higher PsP (56.6) than those who did not report taking anything (48.27). These results indicate that PsP offers a more accurate estimate of arterial elasticity than systolic or diastolic blood pressure, especially in medicated older adults. Results from analyses using systolic and diastolic blood pressures as predictors of WMC are reported in the Supplementary Materials.

The “gold standard” measure for physical fitness is VO₂ max, which is a measure of the maximum oxygen consumption in an individual’s body obtained during a physical exhaustion test. However, older and low-fit individuals may have conditions that prevent them from participating in the physical exhaustion test. In this study, the metabolic equivalent (MET) of VO₂ max was estimated based on the following equation:

$${\displaystyle \begin{array}{ll}{\displaystyle \mathrm{M}\mathrm{E}\mathrm{T}=}& \mathrm{S}\mathrm{e}\mathrm{x}\times 2.77-\mathrm{a}\mathrm{g}\mathrm{e}\times 0.1-\mathrm{b}\mathrm{o}\mathrm{d}\mathrm{y}\mathrm{m}\mathrm{a}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}\\ & {\displaystyle \times 0.17-\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{h}\mathrm{e}\mathrm{a}\mathrm{r}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e}}\\ & {\displaystyle \times 0.03+\mathrm{p}\mathrm{h}\mathrm{y}\mathrm{s}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{l}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{i}\mathrm{t}\mathrm{y}\mathrm{s}\mathrm{c}\mathrm{a}\mathrm{l}\mathrm{e}+18.07}\end{array}}$$

The physical activity scale was measured based on the Physical Activity Scale for the Elderly questionnaire. This measure utilizes easily acquired parameters that are highly predictive of VO₂ max (Jurca et al., 2005). It has been demonstrated to approximate VO₂ max with good accuracy in a large sample (N > 10,000), and recently, Mailey et al. (2010) extended the validity of the MET estimate specifically to older adults, ranging in ages from 60 to 80. McAuley et al. (2011) further validated this measure for smaller sample sizes by showing that there was no significant difference between this estimated measure of MET and the gold-standard, physician-supervised, maximal exercise test in a sample of 86 older adults. Finally, in a survey of more than 32,000 individuals ranging in age from 35 to 70 years, Stamatakis et al. (2013) found MET to be a good predictor of cardiovascular (and overall) mortality, comparable to associations between exercise testing cardiovascular fitness and mortality. High MET indicates higher physical fitness.

Assessments of WMC Using Complex Span Tasks

Computation span

The computation span task is a well-established complex span task that measures WMC (Conway et al., 2005). In this task, participants saw a simple addition or subtraction problem (e.g., 5 + 3 = 7) and indicated whether the arithmetic problem is correct (“p” key) or wrong (“q” key). Participants then recalled the second number of the equation by entering a 3. There were two additional trials involving only one equation. The number of equations, indicating the span length, increased by one after every three trials, with a total of seven possible span lengths. If the order of recalled digits was correct, it was scored as a correct trial. The task would terminate if the participant failed to correctly recall the order of two subsequent trials of the same span length. The final span length of correctly recalled digits indicated the participant’s WMC.

Reading span

The reading span task is also a common complex span used to measure WMC (Daneman & Carpenter, 1980). In this task, participants saw a sentence on the screen (e.g., “a part of the body used for seeing things is the eye”) and indicated whether the sentence was true (“p” key) or false (“q” key). Participants then typed in the last word of the sentence (e.g., eye). There were two additional trials involving only one sentence. The number of sentences increased by one every three trials, with a total of seven possible span lengths. If the order of recalled words was correct, the trial was scored as a correct trial. The task would terminate if the participant failed to recall the order of two trials correctly of the same span length. The final span length of correctly recalled words indicated the participant’s WMC.

A composite score of WMC was calculated by first standardizing the outcomes (z score) from computation span and reading span and then adding the standardized scores. This composite score was used as a proxy of older adults’ WMC in all analyses. Higher composite score indicates a larger WMC.

Results

Table 1 reports the statistics for PsP, MET, and main demographic variables for our healthy aging sample. The 89 participants had an average PsP of 52, ranging from 24 to 97, and an average MET of 7.08, ranging from 2.62 to 12.53 (Table 1). Like VO₂ max, the MET score is biased toward male participants (M_male = 9.27, M_female = 5.97, t(87) = 8.88, p < .01). Age of participants has traditionally been associated with cardiovascular conditions in older adults. In the current sample, age of participants was significantly correlated with PsP (r = 0.21, p = .05), but not with MET. Sex, age, and years of education of participants were controlled for in all analyses.

Table 1.

Demographic Information of Participants Included in Experiment 3

N (M/F)	89 (30/59)
Age (range; SD)	69 (60–92; 8)
Sys. BP (range; SD)	132.1 (105–168; 16.09)
Dia. BP (range; SD)	80.2 (60–109; 10.18)
PsP (range; SD)	51.84 (24–97; 14.75)
MET (range; SD)	7.1 (2.6–12.53; 2.28)
BMI (range; SD)	26.29 (18.2–40.1; 5.1)
RHR (range; SD)	71.55 (50–118; 12.51)
PAS (range; SD)	1.6 (0–3; 1.18)
MoCA (range; SD)	27.73 (24–30; 1.78)
Years of education (range; SD)	16.21 (12–24; 2.43)
GDS (range; SD)	2.69 (0–4; 2.13)
Computation span (range; SD)	3.54 (1–7; 2.04)
Reading span (range; SD)	2.93 (1–7; 1.21)

Notes: SD = standard deviation; Sys. BP = systolic blood pressure; Dia. BP = diastolic blood pressure; PsP = pulse pressure; MET = metabolic equivalent of VO₂ max; BMI = body mass index; RHR = resting heart rate; PAS = Physical Activity Scale for Elderly; MoCA = Montreal cognitive assessment; GDS = Geriatric Depression Scale. Means and ranges of variables are included in the analyses.

Regarding WMC, on average, the current participants had a computation span of 3.54 and a reading span of 2.93. These spans are comparable to those reported in a meta-analysis by Bopp and Verhaeghen (2005; mean computation span: 3.04, mean reading span: 2.86).

Twelve participants had MoCA scores below 26 (24–25), and 38 participants reported taking medications (or supplements) for blood pressure. Additional analyses were conducted to investigate if the MoCA scores and medication status influenced the health–WMC relationships (Supplementary Materials). These analyses included conducting regressions without the 12 low MoCA participants (those with scores <26) and by using MoCA and medication status as additional covariates. Results from these additional analyses are not different from the main analyses, indicating that MoCA and medication status did not influence the relationships between the cardiovascular health factors and WMC. Moreover, older adults taking medication did not differ in terms of WMC from those who reported not taking any medication (t = 0.07, p = .9).

Separate and Combined Effects of Arterial Elasticity and Physical Fitness on WMC

Two step-wise regressions (Table 2) were conducted to assess separate effects of arterial elasticity (Model 1) and fitness (Model 2) on WMC, after controlling for the covariates (i.e., age, sex, and education).

Table 2.

Step-Wise Regression With WMC as the Dependent Variable to Test the Separate Effect of PsP on WMC and Whether the Combined Effects of PsP and MET Could Explain More Variance in WMC Than PsP or MET Alone

Step	IV	Model R2	∆R 2 (from the previous step)	Standardized β of IV	Unstandardized β of IV (SE)
1	Age	0.01		−0.14	−0.03 (0.02)
	Sex			−0.17	−0.06 (0.39)
	Education			0.18	0.12 (0.08)
Model 1
2	PsP	0.07	0.01*	−0.27*	−0.03 (0.01)*
3	MET	0.09	0.11	0.25	0.19 (0.12)
	PsP			−0.23*	−0.03 (0.01)*
Model 2
2	MET	0.05	0.05*	0.33*	0.25 (0.11)*
3	MET	0.09	0.09*	0.25	0.19 (0.12)
	PsP			−0.22*	−0.03 (0.01)*

Notes: MET = metabolic equivalent of VO₂ max; PsP = pulse pressure; IV = independent variable; WMC = working memory capacity. Model R² presents the variance of WMC explained by IVs entered in each step. ∆R² presents changes in variance explained with additional IV. Significant ∆R² suggests that added IVs are significantly improving the predictive power of the model on WMC. Standardized β of IV indicates the individual predictive power of each IV entered. Unstandardized β of IV estimates the exact change in WMC when the IV increases one unit, while holding all of the other IVs constant.

*Indicates significance at α level 0.05.

The last step of each model assessed the combined effects of elasticity and fitness, after accounting for the separate effects of elasticity or fitness on WMC. If the combined effects of elasticity and fitness could explain a significant amount of variance in WMC over and beyond the amount of variance explained by elasticity or fitness alone, R² change (∆R²) from step 2 to step 3 (Table 2) in the models would be significant.

Separate effect of elasticity or fitness on WMC

The first step-wise regression model with PsP as the independent predictor, WMC as the dependent variable, and age, sex, and education as covariates showed that PsP was a significant predictor of WMC in older adults (β = −0.27, p = .01; Table 2).

The second step-wise regression model with MET as the independent predictor, WMC as the dependent variable, and age, sex, and education as covariates showed that MET was a significant predictor of WMC in older adults (β = 0.33, p = .03; Table 2).

Combined effects of arterial elasticity and physical fitness

The last step of each model (Table 2, Step 3) assessed the combined effects of arterial elasticity and fitness over and beyond the separate effects of arterial elasticity or physical fitness on WMC (Step 2). Results from the first model found that the combined effects of arterial elasticity and fitness did not explain more variance in WMC than arterial elasticity alone (∆R² = 0.02, p = .11).

Results from the second model indicate that the combined effects of arterial elasticity and fitness explained significantly more variance in WMC than fitness alone (∆R² = 0.04, p = .04). After accounting for the effects of PsP, MET was no longer a significant predictor of WMC (β = 0.25, p = .12). These results suggest PsP may mediate the relationship between MET and WMC in the current sample of healthy older adults.

Mediation Effect of PsP on the Relationship Between MET and WMC

Results from the above two step-wise regression models suggest a potential mediation effect of PsP on the relationship between MET and WMC. Moreover, PsP and MET were significantly correlated with each other (r = −0.24, p = .01), after controlling for age, sex, and education. Mediation analysis with MET as the independent variable, WMC as the dependent variable, PsP as the mediator, and covariates was conducted with bootstrapping test (5000 iterations) using the PROCESS macro (Hayes, 2018) in SPSS.

In the mediation analysis (Figure 1), the predictive effect of MET on WMC, ignoring the mediator, was significant, β = 0.25, t(84) = 2.2, p = .03. The predictive effect of MET on PsP, the mediator, was also significant, β = −2.44, t(84) = −2.57, p = .01. Furthermore, the mediator (PsP), controlling for MET, significantly predicted declines in WMC, β = −0.03, t(83) = −2.05, p = .04. However, after controlling for the effects of the mediator (PsP), MET was no longer a significant predictor of WMC, β = 0.19, t(83) = 1.67, p = .1. The indirect effect of MET on WMC through PsP (mediation) had an effect size of 0.06, with 95% confidence interval not including zero (95% CI: 0.01–0.13). The mediator (PsP) could account for roughly 30% of the effect of MET on WMC. These results suggest a full mediation effect of PsP on the relationship between MET and WMC in older adults.

Figure 1.

Mediation analysis showed that the effect of MET on working memory capacity was significantly mediated by PsP. a = effect (β) of MET on PsP, b = effect (β) of PsP on working memory capacity after controlling for MET, c = total effect (β) of MET on working memory capacity before controlling for PsP, and c′ = effect (β) of MET on working memory capacity after controlling for PsP. PsP could account for roughly 30% of the total effect, P_M = 0.29. *Indicates significance at α level 0.05. MET = metabolic equivalent of VO₂ max; PsP = pulse pressure.

Discussion

This study is the first to use complex span tasks in order to investigate the effects of multiple cardiovascular health factors on WMC in older adults. Our results show that both cardiovascular health factors under investigation were separately associated with WMC. That is, higher arterial elasticity (i.e., low PsP) and higher physical fitness (i.e., high MET) were separately associated with higher WMC assessed by the complex span tasks. Moreover, these significant associations between cardiovascular health factors and WMC were not affected by cognitive status (MoCA scores) and medication status of older participants. Taking these results a step further, this study then examined the combined effects of these two health factors (i.e., arterial elasticity and fitness) over and beyond the effects of a single factor alone on WMC. The combined beneficial effect of high arterial elasticity and high fitness was larger than the effect of fitness alone on WMC. However, the combined effects of these two factors on WMC were not significantly greater than the effects of arterial elasticity alone. These results suggest that (a) combined effects of arterial elasticity and fitness were mainly driven by arterial elasticity, and therefore (b) arterial elasticity may be the more influential cardiovascular health factor of the two on WMC in healthy older adults.

Medical professionals and researchers suggest an active lifestyle as the most important and cost-effective method to ameliorate the negative effects of high cardiovascular burden, such as low arterial elasticity (Carpio-Rivera et al., 2016). Additionally, numerous exercise (both aerobic and anaerobic) intervention studies have reported the benefit of increasing physical fitness on cognitive and physical health in older adults (for a recent meta-analysis see Falck et al., 2019). The biological mechanisms through which exercise benefits WMC in humans are not clearly understood, but it is suggested that exercise may manifest gains to cognition in older adults through increased brain volumes and brain functions. Some intervention studies that incorporated neuroimaging measures before and after aerobic fitness training have reported increases in hippocampal volume and increases in frontal activation in older adults (Erickson et al., 2011; Maass et al., 2015; Voss et al., 2010). It is also possible that aerobic and anaerobic exercise benefits cognitive performance in older adults through its relationship with various biological factors in the brain, such as neural growth factors (for a meta-analysis on effects of exercise training on brain derived neurotrophic factor (BDNF) in older adults see Marinus et al., 2019), neuronal metabolite concentration (e.g., N-Acetylaspartate; Erickson et al., 2012), or arterial elasticity as observed in this study.

In this study, physical fitness on its own was predictive of WMC and had higher predictive power (β = 0.33) than arterial elasticity (β = −0.27) on WMC. However, the fitness effect on WMC was rendered insignificant when arterial elasticity was taken into account. This suggests that the two cardiovascular health factors might not be working in parallel to create an additive effect on WMC. Instead, the relationship between fitness and WMC was found to be fully mediated by arterial elasticity. Arterial elasticity accounted for 30% of variance within the relationship between fitness and WMC in older adults. Thus, fitness effects on WMC could be explained by its relationship with arterial elasticity, a biologically based cardiovascular health factor. Similar mediation effects of a biologically based health factor on the relationship between fitness and working memory have been reported in a 2012 study by Erickson et al. (2012). In that study, the relationship between fitness and WMC was partially mediated by the concentration of a neuronal metabolite (NAA) that is deemed important for the viability of neurons in the brain. It is therefore possible that high fitness (VO₂ max) in older adults is associated with high arterial wall elasticity and/or high neuronal metabolite production, which directly benefits older adults’ WMC. Future studies should explore all these potential mediators in order to gain a more complete understanding of the biological mediators of physical fitness and WMC.

Because arterial elasticity drives the combined effects of the two health factors on WMC in older adults, these results suggest that biologically based health factors in older adults (i.e., arterial elasticity) can influence their working memory (i.e., WMC). Past researchers have consistently reported significant associations between larger digit span in older adults with better cardiovascular health (Chang et al., 2013; Elias et al., 2010; Gayda et al., 2017; Ihle et al., 2017; Newson & Kemps, 2008; Raz et al., 2007; Shin et al., 2012). Although no study has examined the effects of both health factors on short-term memory measure of digit span, it is possible to find similar mediation effects of arterial elasticity on the relationship between digit span and fitness. Moreover, for other cognitive abilities, such as episodic memory, executive functions, and reasoning, that are not only associated with working memory but also improve with working memory training in healthy aging (Basak & O’Connell, 2016), it is possible that some of the observed age-related differences in these abilities could be attributed to individual differences in arterial elasticity among older participants. Given the prevalence of low arterial elasticity (e.g., high blood pressure and PsP) among older adults, it is important for future studies on aging and working memory to assess such biologically based health factors, particularly easy-to-obtain cardiovascular health measures (e.g., PsP) in older participants. Much of observed age-related declines in WMC and its related cognitive abilities could be attributed to individual differences in cardiovascular health, particularly arterial elasticity.

Results from our mediation analysis also suggest that arterial elasticity should be a key target for exercise intervention studies in order to achieve long-term benefits in WMC and its related cognition in older adults. A recent meta-analysis (Falck et al., 2019) has reported gains in both physical and cognitive functions in older adults following various exercise training regimes. For aerobic exercise training, cognitive gains might be related to improvements in arterial plasticity (Herrod et al., 2018), although such association has not been specifically investigated for WMC. Individual aerobic exercise studies that reported gains in WMC did not report PsP or blood pressure in older participants (Gothe et al., 2014; Langlois et al., 2013; Voss et al., 2013).

The effects of anaerobic exercise (e.g., strength or weight training, body sculpting, Taichi) on arterial elasticity measures, such as PsP, have been inconsistent. Some studies reported a benefit of anaerobic exercise on blood pressure in older adults (Jessup et al., 1998; Ladawan et al., 2017), while others have reported worsening arterial elasticity after anaerobic exercise (Bertovic et al., 1999; Kingwell, 2002; Lu et al., 2013). Given the mixed findings regarding anaerobic exercises and arterial elasticity, it is possible that anaerobic exercise benefits cognition in older adults through other biological mechanisms, such as BDNF (Marinus et al., 2019), not through arterial elasticity.

Physicians, in addition to prescribing blood control medications, often would encourage older adults to exercise regularly, sometimes based on the strong predictive power of fitness on cognition from studies that only measured physical fitness. However, many older adults find daily exercise difficult to maintain, because of increased frailty or lack of access to senior-oriented exercise programs (Schutzer & Graves, 2004). The situation has been exacerbated for the past year with the coronavirus disease 2019 pandemic, when older adults are advised to stay home (Jernigan, 2020). Results from this study suggest that maintaining healthy PsP through diet or medication or supplements is critical for maintaining cognitive health in the aging population.

A potential limiting factor of this study could be that all older adults in the sample were at least high school educated—this number may be slightly higher than the national average. Although a 2018 report from U.S. Census Bureau (Current Population Survey, Annual Social and Economic Supplement) reports that 87% of adults older than the age of 65 have completed high school education. This slight difference can be attributed to the urban/suburban population of Collin and Dallas counties, where 2019 estimates of high school-educated adults are 94% and 90%, respectively (census.gov). Urban populations are different than those drawn from rural communities where education levels are typically lower (Marré, 2017). Moreover, to reduce the risk of including nonnormative aging in our sample, we recruit independently living older adults through flyers placed in the community and recreational centers—facilities more accessible to older adults with higher education (Bethancourt et al., 2014). It is important to note that the relationships between cardiovascular health factors and WMC in this study were independent of educational attainment. Because education is considered to be a protective factor that helps build a cognitive reserve in older adults (Wilson et al., 2019) and worsening cardiovascular health factors counteract such cognitive benefits, the results from our sample of relatively well-educated and physically healthy (living independently) older adults emphasize the importance of evaluating the impact of multiple cardiovascular health factors on working memory in healthy aging.

Funding

This work was supported by a grant from the National Institutes of Health (titled “Strategic Training to Optimize Neurocognitive Functions in Older Adults” under award number R56AG060052) and a grant from Advanced Imaging Research Center at UT Southwestern Medical Center to C. Basak. This work was also partially supported by a Post Graduate Scholarship from the Natural Science and Engineer Council to S. Qin.

Conflict of Interest

None declared.