Introduction
Normal aging may result in structural and functional modifications to the brain, such as a reduction in brain volume (Good et al., Reference Good, Johnsrude, Ashburner, Henson, Friston and Frackowiak2001; Peelle, Cusack, & Henson, Reference Peelle, Cusack and Henson2012), decline in resting cerebral blood flow (CBF) (Brown et al., Reference Brown, McMorris, Longman, Leigh, Hill, Friedenreich and Poulin2010; Zimmerman et al., Reference Zimmerman, Sutton, Low, Fletcher, Tan, Schneider-Garces and Fabiani2014), and cerebrovascular reactivity (Barnes, Taylor, Kluck, Johnson, & Joyner, Reference Barnes, Taylor, Kluck, Johnson and Joyner2013; Brown et al., Reference Brown, McMorris, Longman, Leigh, Hill, Friedenreich and Poulin2010). These changes in cerebrovascular health and cell integrity have been associated with an increased risk of stroke (Gupta et al., Reference Gupta, Chazen, Hartman, Delgado, Anumula, Shao and Sanelli2012), premature mortality (Portegies, de Bruijn, Hofman, Koudstaal, & Ikram, Reference Portegies, de Bruijn, Hofman, Koudstaal and Ikram2014; Sabayan et al., Reference Sabayan, van der Grond, Westendorp, Jukema, Ford, Buckley and de Craen2013), Alzheimer’s disease and related dementias (Amieva et al., Reference Amieva, Jacqmin-Gadda, Orgogozo, Le Carret, Helmer, Letenneur and Dartigues2005; Lautenschlager, Cox, & Cyarto, Reference Lautenschlager, Cox and Cyarto2012), and declining cognitive function (Davenport, Hogan, Eskes, Longman, & Poulin, Reference Davenport, Hogan, Eskes, Longman and Poulin2012; Matteis, Troisi, Monaldo, Caltagirone, & Silvestrini, Reference Matteis, Troisi, Monaldo, Caltagirone and Silvestrini1998). Physical activity (PA) has been identified as one of the most promising modifiable lifestyle factors for improving cerebrovascular health (Ainslie et al., Reference Ainslie, Cotter, George, Lucas, Murrell, Shave and Atkinson2008; Bailey et al., Reference Bailey, Marley, Brugniaux, Hodson, New, Ogoh and Ainslie2013; Brown et al., Reference Brown, McMorris, Longman, Leigh, Hill, Friedenreich and Poulin2010; Chapman et al., Reference Chapman, Aslan, Spence, Defina, Keebler, Didehbani and Lu2013; Prakash, Voss, Erickson, & Kramer, Reference Prakash, Voss, Erickson and Kramer2015; Tarumi et al., Reference Tarumi, Gonzales, Fallow, Nualnim, Pyron, Tanaka and Haley2013) and preventing age-associated cognitive decline (Abbott et al., Reference Abbott, White, Ross, Masaki, Curb and Petrovitch2004; Buchman et al., Reference Buchman, Boyle, Yu, Shah, Wilson and Bennett2012; Laurin, Verreault, Lindsay, MacPherson, & Rockwood, Reference Laurin, Verreault, Lindsay, MacPherson and Rockwood2001; Rovio et al., Reference Rovio, Kareholt, Helkala, Viitanen, Winblad, Tuomilehto and Kivipelto2005; Scarmeas et al., Reference Scarmeas, Luchsinger, Schupf, Brickman, Cosentino, Tang and Stern2009; Tarumi et al., Reference Tarumi, Gonzales, Fallow, Nualnim, Pyron, Tanaka and Haley2013). Furthermore, the benefits of PA include increased cerebrovascular function (Ainslie et al., Reference Ainslie, Cotter, George, Lucas, Murrell, Shave and Atkinson2008; Bailey et al., Reference Bailey, Marley, Brugniaux, Hodson, New, Ogoh and Ainslie2013; Brown et al., Reference Brown, McMorris, Longman, Leigh, Hill, Friedenreich and Poulin2010; Burdette et al., Reference Burdette, Laurienti, Espeland, Morgan, Telesford, Vechlekar and Rejeski2010; Chapman et al., Reference Chapman, Aslan, Spence, Defina, Keebler, Didehbani and Lu2013; Davenport et al., Reference Davenport, Hogan, Eskes, Longman and Poulin2012), decreased risk of cerebrovascular and cardiovascular diseases (Qiu & Fratiglioni, Reference Qiu and Fratiglioni2015) and a decreased risk of Alzheimer’s disease and related dementias (de Bruijn et al., Reference de Bruijn, Schrijvers, de Groot, Witteman, Hofman, Franco and Ikram2013; Nation et al., Reference Nation, Wierenga, Clark, Dev, Stricker, Jak and Bondi2013; Rovio et al., Reference Rovio, Kareholt, Helkala, Viitanen, Winblad, Tuomilehto and Kivipelto2005).
Previous research indicates that older adults who engage in more physically active recreational activities or have higher cardiovascular fitness are at lower risk for cognitive decline compared to inactive older adults (Blondell, Hammersley-Mather, & Veerman, Reference Blondell, Hammersley-Mather and Veerman2014; Colcombe et al., Reference Colcombe, Kramer, Erickson, Scalf, McAuley, Cohen and Elavsky2004; Forbes, Thiessen, Blake, Forbes, & Forbes, Reference Forbes, Thiessen, Blake, Forbes and Forbes2013; Kramer et al., Reference Kramer, Hahn, Cohen, Banich, McAuley, Harrison and Colcombe1999; Prakash et al., Reference Prakash, Voss, Erickson and Kramer2015, Reference Prakash, Voss, Erickson, Lewis, Chaddock, Malkowski and Kramer2011). Attention has been focused on the effects of current leisure time or recreational PA on cognitive function, demonstrating the potential importance of remaining active in older age (Rovio et al., Reference Rovio, Kareholt, Helkala, Viitanen, Winblad, Tuomilehto and Kivipelto2005; Weuve et al., Reference Weuve, Kang, Manson, Breteler, Ware and Grodstein2004; Yaffe, Barnes, Nevitt, Lui, & Covinsky, Reference Yaffe, Barnes, Nevitt, Lui and Covinsky2001). For example, in females ≥65 years, walking more than three times per week was associated with a decreased risk of cognitive decline (Yaffe et al., Reference Yaffe, Barnes, Nevitt, Lui and Covinsky2001). Rovio et al. (Reference Rovio, Kareholt, Helkala, Viitanen, Winblad, Tuomilehto and Kivipelto2005) demonstrated that leisure-time PA in midlife (age 44–57 years) is associated with a decreased risk of dementia in late life, while Verghese et al. (Reference Verghese, Lipton, Katz, Hall, Derby, Kuslansky and Buschke2003) showed increased participation in leisure activities for individuals ≥75 years old reduced the risk of developing dementia and Alzheimer’s disease.
Physical activity can be measured using metabolic equivalents (METs). A MET is defined as the ratio of the associated metabolic rate for an activity compared to the resting metabolic rate. One MET is equivalent to the amount of oxygen that is metabolized while at rest in a seated position, and is approximately 3.5 mL/kg/min of oxygen consumption (
$\dot{\rm V}{\rm O}_{2} $
; normalized to body mass) (Jette, Sidney, & Blumchen, Reference Jette, Sidney and Blumchen1990). With increasing intensity of PA, the MET value increases; for example, a 3 MET activity is achieved by walking at 2.5 miles per hour on a flat firm surface, while cross country hiking is rated as a 6 MET activity (Ainsworth et al., Reference Ainsworth, Haskell, Herrmann, Meckes, Bassett, Tudor-Locke and Leon2011). It has been demonstrated that increasing MET-hr/week spent in recreational and household activities over the past 2 weeks is associated with a lower risk of dementia (de Bruijn et al., Reference de Bruijn, Schrijvers, de Groot, Witteman, Hofman, Franco and Ikram2013), and increasing MET-hr/week of leisure time PA over the past year resulted in increased mean cognitive scores (Weuve et al., Reference Weuve, Kang, Manson, Breteler, Ware and Grodstein2004).
Cerebrovascular health and function can be measured using indices of resting CBF and cerebrovascular reactivity. Cerebrovascular reactivity is the magnitude of change in CBF for a given stimulus [e.g., increased partial pressure of CO2 (PCO2)], with enhanced vascular reactivity thought to represent better cerebrovascular function and health (Brown et al., Reference Brown, McMorris, Longman, Leigh, Hill, Friedenreich and Poulin2010; Davenport et al., Reference Davenport, Hogan, Eskes, Longman and Poulin2012). Previous research provides evidence suggesting that increased levels of PA or cardiovascular fitness are associated with improved cerebrovascular functioning (Ainslie et al., Reference Ainslie, Cotter, George, Lucas, Murrell, Shave and Atkinson2008; Bailey et al., Reference Bailey, Marley, Brugniaux, Hodson, New, Ogoh and Ainslie2013; Burdette et al., Reference Burdette, Laurienti, Espeland, Morgan, Telesford, Vechlekar and Rejeski2010), with additional studies indicating that cerebrovascular functioning is also associated with improved neurocognitive function within the same sample (Brown et al., Reference Brown, McMorris, Longman, Leigh, Hill, Friedenreich and Poulin2010; Chapman et al., Reference Chapman, Aslan, Spence, Defina, Keebler, Didehbani and Lu2013; Tarumi et al., Reference Tarumi, Gonzales, Fallow, Nualnim, Pyron, Tanaka and Haley2013). Reduced cerebrovascular functioning has been proposed as a mechanism underlying cognitive and cerebrovascular dysfunction (Davenport et al., Reference Davenport, Hogan, Eskes, Longman and Poulin2012). Recently it has been shown that indices of cerebrovascular function mediate the relation between PA and cognition in healthy young adults providing support for the hypothesis that cerebrovascular health is a plausible pathway linking frequent PA and improved cognitive status (Guiney, Lucas, Cotter, & Machado, Reference Guiney, Lucas, Cotter and Machado2015). Specifically, more frequent PA was associated with enhanced cognitive control and this relationship was mediated by cerebrovascular reactivity to PCO2, as per statistical mediation analysis (Guiney et al., Reference Guiney, Lucas, Cotter and Machado2015). Together these studies suggest that cardiovascular and cerebrovascular health may contribute to better cognitive functioning.
Dysfunction in both cognitive performance (Andel et al., Reference Andel, Crowe, Pedersen, Fratiglioni, Johansson and Gatz2008; Rovio et al., Reference Rovio, Kareholt, Helkala, Viitanen, Winblad, Tuomilehto and Kivipelto2005; Yaffe et al., Reference Yaffe, Barnes, Nevitt, Lui and Covinsky2001) and cerebrovascular functioning (Tarumi et al., Reference Tarumi, Gonzales, Fallow, Nualnim, Pyron, Tanaka and Haley2013) can begin in midlife (i.e., ~40–60 years old) rather than exclusively in old age (Kareholt, Lennartsson, Gatz, & Parker, Reference Kareholt, Lennartsson, Gatz and Parker2011). Previous research has highlighted the need for additional investigations addressing the influence of PA on cognitive abilities throughout the lifespan to complement current work that focuses attention on periods more proximate to the onset of disease. Our study addresses this knowledge gap regarding the effects of lifetime total PA on cognitive and cerebrovascular outcomes in mid to later life while examining potential sex differences in these outcomes. Sex differences were considered as there is evidence suggesting that males and females have different physiological responses to PA (Brown, Peiffer, & Martins, Reference Brown, Peiffer and Martins2013; Colcombe & Kramer, Reference Colcombe and Kramer2003; Ho, Woo, Sham, Chan, & Yu, Reference Ho, Woo, Sham, Chan and Yu2001; Laurin et al., Reference Laurin, Verreault, Lindsay, MacPherson and Rockwood2001). A variety of mixed sex studies have found that exercise positively impacts both sexes; however, the effects of increased exercise on cognitive functioning seem to be more pronounced in women (Brown et al., Reference Brown, Peiffer and Martins2013; Ho et al., Reference Ho, Woo, Sham, Chan and Yu2001; Laurin et al., Reference Laurin, Verreault, Lindsay, MacPherson and Rockwood2001). Additionally, when studying cognition in a mixed-sex study it is important to account for sex differences (Kimura, Reference Kimura2002; Kimura & Hampson, Reference Kimura and Hampson1994).
This study tests the mediating effect of current cerebrovascular health on the relation between measures of lifetime PA and current cognitive functioning. We hypothesized that increased lifelong PA is associated with better global cognition and cerebrovascular function is associated with current cognitive performance, completing the proposed mediation relationship.
Methods
Research Participants
The study population comprised participants in an ongoing intervention-cohort study on the effects of a 6-month aerobic exercise intervention on cerebrovascular regulation and cognitive function in middle-aged and older adults. The Brain in Motion study methods and protocol for the eligibility screening process has been previously described (Tyndall et al., Reference Tyndall, Davenport, Wilson, Burek, Arsenault-Lapierre, Haley and Poulin2013). Participants (n=264) are healthy, community dwelling, currently inactive, middle-aged and older adults and were recruited through the use of media, posters and advertisements in local newspapers, communities, and through the University of Calgary. Eligibility criteria included: English speaking males or females ≥55 years old, considered currently inactive (<30 min of moderate exercise 4 days/week or 20 continuous min of vigorous exercise two days/week), able to walk independently outside and up and down at least 20 stairs, a body mass index (BMI) less than 35 kg/m2 (to avoid co-morbidities associated with obesity), no history of clinically active cardiovascular disease or obstructive airway disease, non-smoker for the past 12 months, no major trauma or surgery in the last 6 months, no debilitating neurological disorders, physician clearance, and written informed consent. A detailed flow of participants is provided in Figure 1. The sample consists of participants who completed the Lifetime Total Physical Activity Questionnaire (LTPAQ) in addition to the neuropsychological assessment, and the cerebrovascular blood flow test (n=226) at baseline. The University of Calgary Conjoint Health Research Ethics Board approved all study procedures.
Fig. 1 Participant flow for the Brain in Motion Study, Calgary, Alberta, Canada.
Assessment of Lifetime Physical Activity
Lifetime total PA was assessed using the interviewer-administered LTPAQ, a tool with demonstrated reliability (Friedenreich, Courneya, & Bryant, Reference Friedenreich, Courneya and Bryant1998). This questionnaire assesses occupational, transportation, household and recreational physical activities from childhood to time of interview. Additionally, the frequency, duration, and intensity of PA are also reported. Before the interview, participants received two recall calendars as memory aids to complete. An interviewer trained in cognitive interviewing techniques used the calendars to help facilitate recall of PA history.
Lifetime total PA was the main variable of interest for this study. Intensities were assessed in two ways: (1) by self-report as sedentary (only for occupational activity described as activities sitting down), light (activities done mainly standing that do not increase heart rate and cause no sweating), moderate (activities that cause heart rate to increase slightly and cause light sweating), and vigorous activity (activities that cause heart rate to increase substantially and cause heavy sweating) and (2) as assigned by the study staff. For the latter approach, a MET value for each reported activity was assigned based on the Compendium of Physical Activities (Ainsworth et al., Reference Ainsworth, Haskell, Herrmann, Meckes, Bassett, Tudor-Locke and Leon2011). The main predictor for these analyses was the average MET-hours per week per year of life (MET-hr/week/year) of lifetime PA estimated as the sum of occupational, transportation, household, and recreational activity done from childhood to time of LTPAQ. For this study, additional analyses by type of activity, intensity of activity (0–3 METs: low, 3–6 METs: moderate, and >6 METs: vigorous) and activity during different age period in life (ages <20, 21–35, 36–50, 51–65 years) were conducted.
Neuropsychological Assessment
This assessment consisted of a 2.5 hour neuropsychological test battery administered by trained staff. The test battery was composed of eleven tests assessing seven cognitive domains including verbal memory, figural memory, processing speed, executive functioning, complex attention, verbal knowledge, and spatial reasoning. The list of neuropsychological tests corresponding to these domains of cognitive activity are displayed in Supplementary Table 1, and a detailed description has been previously published (Tyndall et al., Reference Tyndall, Davenport, Wilson, Burek, Arsenault-Lapierre, Haley and Poulin2013). Seven cognitive domain scores were calculated by taking the average Z-score of all tests within each domain. The global cognition score is the sum of seven equally weighted domain Z-scores, used as the cognitive outcome for all analyses. At the time of the cognitive assessment, participants also completed the North American Adult Reading Test (NAART) as a measure of premorbid verbal intellectual ability (Blair & Spreen, Reference Blair and Spreen1989). The NAART is sensitive to education (both formal and informal) and insensitive to mild forms of cognitive impairment, making it a better covariate than formal education when assessing cognitive abilities (Uttl, Reference Uttl2002).
Assessment of Indices of Cerebrovascular Blood Flow
Participants underwent a 2-hour assessment administered by trained staff. Two hours before testing, participants fasted and refrained from exercising. Blood flow velocity of the middle cerebral artery (MCAv) was measured using a 2-MHz pulsed transcranial Doppler ultrasound system recording measurements at an optimal placement slightly above and in front of the right ear (Aaslid, Markwalder, & Nornes, Reference Aaslid, Markwalder and Nornes1982; Poulin, Liang, & Robbins, Reference Poulin, Liang and Robbins1996; Poulin & Robbins, Reference Poulin and Robbins1996). Peak MCAv, heart rate, beat-by-beat blood pressure measurements, and arterial O2 saturation were measured continuously throughout the protocol. Dedicated software recorded the exhaled CO2 and O2 at the end of each breath (referred to as end-tidal PCO2 and PO2) during 10 min of seated rest. Each participant had his/her nose occluded with a nose clip and breathed room air through a mouthpiece. A fine capillary line inserted in a port immediately distal to the mouthpiece and connected to a mass spectrometer measured the concentration of CO2 and O2 continuously at the mouth, and breath-by-breath values for end tidal CO2 (PETCO2) and O2 (PETO2) were determined. These end-tidal values were averaged over the 10 min of rest and were used to determine the desired PETCO2 and PETO2 to assess the cerebrovascular response to the changes in the pressure of CO2, also referred to as euoxic hypercapnia testing. The euoxic hypercapnia test lasted 12 min and included two 3-min step increases in PETCO2 as previously described (Brown et al., Reference Brown, McMorris, Longman, Leigh, Hill, Friedenreich and Poulin2010; Tyndall et al., Reference Tyndall, Davenport, Wilson, Burek, Arsenault-Lapierre, Haley and Poulin2013). Physiological responses were calculated as the mean responses over the final 30 s of each stage during the hypercapnic challenges. A more technical description of the testing protocol can be found in the supplementary material.
This protocol yielded four measures of cerebrovascular function for analysis, including peak velocity of blood moving through the MCA (
$\bar{\rm V}{\rm P}$
), cerebrovascular conductance (CVC; MCAv/Mean Arterial Pressure),
$\bar{\rm V}{\rm P}$
, and CVC reactivity during the hypercapnic challenge. Specifically
$\bar{\rm V}{\rm P}$
reactivity was calculated as the change in
$\bar{\rm V}{\rm P}$
divided by the change in PETCO2 from +1 to +8 mmHg, while CVC reactivity is the change in CVC divided by the change in PETCO2 from +1 to +8 mmHg. These measures are widely used in the cerebrovascular literature using transcranial Doppler ultrasound techniques (Aengevaeren, Claassen, Levine, & Zhang, Reference Aengevaeren, Claassen, Levine and Zhang2013; Ainslie et al., Reference Ainslie, Cotter, George, Lucas, Murrell, Shave and Atkinson2008; Bailey et al., Reference Bailey, Marley, Brugniaux, Hodson, New, Ogoh and Ainslie2013; Barnes et al., Reference Barnes, Taylor, Kluck, Johnson and Joyner2013; Brown et al., Reference Brown, McMorris, Longman, Leigh, Hill, Friedenreich and Poulin2010; Demirkaya, Uluc, Bek, & Vural, Reference Demirkaya, Uluc, Bek and Vural2008; Murrell et al., Reference Murrell, Cotter, Thomas, Lucas, Williams and Ainslie2013; Zhu et al., Reference Zhu, Tarumi, Tseng, Palmer, Levine and Zhang2013).
Additional Measures
At baseline socio-demographic, health and lifestyle, maximal aerobic capacity (
$\dot{\rm V}{\rm O}_{2} $
max), and anthropometric were obtained (Table 1). Health and lifestyle information was obtained through self-reported questionnaires and included mood, alcohol consumption, and dietary intake assessed with the Canadian Diet History Questionnaire I (DHQI) (Csizmadi et al., Reference Csizmadi, Kahle, Ullman, Dawe, Zimmerman, Friedenreich and Subar2007); hypertensive status (based on resting blood pressure measures and medications reported); and smoking history. Maximal aerobic capacity was obtained using a motorized treadmill following the Bruce protocol (Paterson, Cunningham, Koval, & St Croix, Reference Paterson, Cunningham, Koval and St Croix1999) described elsewhere (Tyndall et al., Reference Tyndall, Davenport, Wilson, Burek, Arsenault-Lapierre, Haley and Poulin2013). Anthropometric measures were taken by trained staff and included height, weight, BMI, percent body fat (obtained from bioelectrical impedance analysis), and waist circumference. For participants who provided additional genetic consent, a blood sample was taken for genetic testing that included APOE ε4 genotyping.
Table 1 Baseline characteristics for participants (n=226) in the Brain in Motion Study, Calgary, Alberta, Canada
Note. T-statistics for continuous data and Chi2statistics for categorical data.
a All p-values are comparing differences between males and females.
NAART=North American Adult Reading Test;
$\dot{\rm V}{\rm O}_{2} $
max=Maximal oxygen uptake; BMI=Body mass index; APOE ε4=Apolipoprotein ε4 genotype;
$\bar{\rm V}{\rm P}$
=blood flow velocity at +1 mmHg; CVC=cerebrovascular conductance at +1 mmHg;
$\bar{\rm V}{\rm P}$
reactivity=cerebral blood flow reactivity to a hypercapnic challenge from +1 mmHg to +8 mmHg; CVC reactivity=cerebrovascular conductance reactivity to a hypercapnic challenge from +1 mmHg to +8 mmHg.
Statistical Analyses
Descriptive statistics were prepared to characterize the study population and to examine differences between sexes. Continuous variables were summarized using means and standard deviations, while frequency distributions were used for categorical variables. Chi square tests were used to identify between-group differences for categorical variables and a Spearman’s correlation was used to assess potential collinearity among predictor variables. The global cognition and sub-domain Z-scores were calculated from raw data. Both cognitive and cerebrovascular outcome measures were assessed for normality using the Shapiro-Wilk test (Shapiro & Wilk, Reference Shapiro and Wilk1965). Continuous variables considered as confounders were age, NAART, mood,
$\dot{\rm V}{\rm O}_{2} $
max, blood pressure, BMI, percent body fat, waist circumference, waist to hip ratio, cholesterol, HDL, LDL, total HDL, triglycerides, fasting glucose, alcohol consumption, and calories consumed/day. While the categorical variables considered were sex, marital status, income, retirement status, education, hypertensive status, hypercholesterolemia, smoking status, and APOE ε4 genotype. Multiple robust linear regression analyses were used with all final lifetime PA models using lifetime PA predictors adjusting for age, sex, NAART,
$\dot{\rm V}{\rm O}_{2} $
max, BMI, and interaction terms (age-sex, age-predictor, sex-predictor, age-sex-predictor). Final covariates were chosen using both stepwise regression and assessment of the coefficient of determination (model R2). All other variables were disregarded, since they did not improve the fit of the model. In subsequent analyses looking at the type, intensity and life period of PA, each respective grouping of activity was also controlled for (e.g., model for recreational PA adjusts for occupational and household activity). To characterize the relation between lifetime PA and cognitive functioning, while properly addressing the potential mediating effects of cerebrovascular indices (CVC,
$\bar{\rm V}{\rm P}$
, CVC reactivity and
$\bar{\rm V}{\rm P}$
reactivity), mediation analysis was used (Baron & Kenny, Reference Baron and Kenny1986). To test the hypothesis that cerebrovascular function mediates the lifetime PA-cognitive functioning relation, the results of robust linear regression were assessed to determine if statistical significance (α=0.05) was achieved. The following inter-relations required for mediation were assessed: the predictor lifetime PA had to be significantly associated with both the outcome cognitive functioning and the mediator cerebrovascular regulation, and cerebrovascular regulation had to be significantly associated with cognitive functioning. If these relations were statistically significant, the Sobel’s test was used to determine the significance of the indirect mediating effects, or the amount of mediation present (Baron & Kenny, Reference Baron and Kenny1986). Figure 2 represents the analytical framework for mediation analysis highlighting relevant regression coefficients.
Fig. 2 Analytical framework for mediation analysis.Note:
$\bar{\rm V}{\rm P}$
was used as a representative measure of cerebrovascular function in this model.
Two post hoc mediation analyses were performed assessing the relation between past year PA and current fitness on cognitive performance, assessing the mediating effect of cerebrovascular function to determine if current PA levels or fitness are more important for cerebrovascular health and cognitive function than lifetime exposure. All predictors were evaluated for statistical significance at α=0.05. To assess the fit of each model, R2 was used to measure the proportion of variance in the dependent variable that is explained by the robust linear model. To adjust for multiple regression comparisons, a Bonferroni correction was calculated for the analyses performed for type, intensity, and timing of PA if the main relation of interest was significant (Figure 3). No corrections were calculated for the between-group differences in Tables 1, 2, and 3, as these were for descriptive purposes. All statistical analyses were performed in STATA 13.1 (StataCorp, 2013).
Fig. 3 Methodological structure for determining multiple regression adjustments. The main relationship had to be significant at 0.05 to proceed with any other analyses.
Table 2 Average lifetime physical activity measures for male and female participants (n=226) in the Brain in Motion Study, Calgary, Alberta, Canada
a All p-values are comparing differences between males and females.
MET(s)=metabolic equivalents.
Table 3 Result of robust regression for the association between lifetime physical activity and global cognition, including all predictors that were adjusted for
Note. Multivariable adjusted for age at the time of LTPAQ interview, sex, NAART,
$\dot{\rm V}{\rm O}_{2} $
max, BMI, age-sex, age-lifetime PA, sex-lifetime PA, and age-sex-lifetime PA.
R2 attributable to lifetime PA=18.4%, Adjusted model R2=34.7%.
*p<.05 for the overall relationship between lifetime PA and global cognitive performance.
PA=physical activity; NAART=North American Adult Reading Test;
$\dot{\rm V}{\rm O}_{2} $
max=maximal aerobic capacity; BMI=body mass index.
Results
Demographics
Participants had a mean age of 66.5±6.4 years (n=226; 118 females) on study entry, were well educated with a moderate to high socioeconomic status, and 55.8% were retired (Table 1). Table 1 provides information on demographic, health, genetic, lifestyle, and cerebrovascular measures. The distributions of the APOE ε4 allele are similar to those found in the general population; 25.9% of participants were APOE ε4+ while 74.1% were APOE ε4- (McKay et al., Reference McKay, Silvestri, Chakravarthy, Dasari, Fritsche, Weber and Patterson2011). Descriptive statistics for lifetime PA are presented in Table 2. As expected, the neuropsychological raw test scores had few absolute sex differences (Kimura & Hampson, Reference Kimura and Hampson1994), but the female advantages on verbal tests are consistent with known sex effects on memory and verbal fluency (Bleecker, Bolla-Wilson, Agnew, & Meyers, Reference Bleecker, Bolla-Wilson, Agnew and Meyers1988; Kramer, Delis, & Daniel, Reference Kramer, Delis and Daniel1988; Weiss, Kemmler, Deisenhammer, Fleischhacker, & Delazer, Reference Weiss, Kemmler, Deisenhammer, Fleischhacker and Delazer2003). See Supplementary Table 1.
Lifetime PA and Current Cognitive Function
Cognitive and cerebrovascular outcomes were not normally distributed. Since no transformations resolved the issue with non-normal distributions, robust linear regression was used. Three participants had missing NAART scores since English was a second language and were excluded from all analyses. Results presented are for the remaining 223 participants (n=116 female). All analyses using measures of lifetime PA controlled for age, sex, NAART,
$\dot{\rm V}{\rm O}_{2} $
max, and interaction terms age-sex, age-lifetime PA, sex-lifetime PA, age-sex-lifetime PA. The adjusted model for the relation between lifetime PA and cognitive functioning is significant (p=.045). With every unit increase in MET-hr/week/year of lifetime total PA, there was a 0.40 increase in global cognition Z-score (Table 3).
Table 4 Adjusted models for the relation between type, intensity and life periods of physical activity and cognition
Note. Multivariable adjusted for age at the time of LTPAQ interview, NAART,
$\dot{\rm V}{\rm O}_{2} $
max, BMI and interaction terms (age–sex, age–lifetime PA predictor, sex–lifetime PA predictor, age–sex–lifetime PA predictor).
*p<0.05.
δ*p<0.005 (Bonferonni corrected p-value).
MET(s)=metabolic equivalent.
Figure 4 describes the significant three-way interaction between age, sex and lifetime PA, using age categories to depict the interaction. This statistically significant interaction indicates that the relation between lifetime PA and global cognition differs by sex and age; for males the amount of lifetime PA decreases with increasing age with the opposite relation observed in females. Figure 5 describes how cognitive performance changes with increasing lifetime PA using age categories (≤65 and >65) for males and females. There was a significant difference between global cognition scores between males and females at all lifetime PA levels (p<.05), and males have increasing cognition score as levels of lifetime PA increase (p<.0001) (Figure 5). Lifetime PA was not associated with any measures of cerebrovascular health, and cerebrovascular measures were not associated with cognition scores (data not shown).
Fig. 4 Interaction between age, sex and lifetime physical activity. *p-value <.05.Note: Age group trichotomized to display interaction, continuous variable used in all analyses.
Fig. 5 Relation between lifetime PA and global cognitive performance. Created using regression coefficients from Table 3. ***p-value<.0001. **p-value<.001. *p-value<.005.
Relative Components of Lifetime PA and Current Cognitive Function
The relation between lifetime PA and global cognition was significant at α=0.05; therefore, subsequent analyses were performed assessing type, intensity, and timing of PA (Figure 3). The results presented for the type, intensity, and timing of PA adjust for the same covariates and interaction terms as the main relation between total lifetime PA and cognition. There were statistically significant relations between lifetime recreational PA, vigorous intensity PA, and PA done between childhood to age 20 and age 21 to age 35. Specifically, for every unit increase in MET-hr/week/year of lifetime recreational PA there was a 1.18 increase in global cognition Z-score (p=.021). For vigorous intensity PA over lifetime, there was a 9.85 increase in global cognition Z-score for every hr/week of activity (p=.004). For every unit increase in MET-hr/week/year of PA from early childhood to age 20, there was a 0.47 increase (p=.036), and for activity between the ages of 21 and 35 years old, there was a 0.36 increase in global cognition Z-score (p<.0001). Table 4 summarizes the results for the associations between type, dose, and timing of lifetime PA and their associations with cerebrovascular indices, if significant. The three-way interaction between age, sex, and predictor was used in these analyses; however, the interaction effect is not shown. These analyses were repeated controlling for all other types of activity, all other intensities, and all other life periods in the same multivariate model (Table 5). These relations remained significant indicating the unique variance of these findings. Physical activity done from age 21 to 35 switched from significant to trending toward significance (p=.053) when controlling for all age periods; however, when just controlling for past year PA, it remained significant (Table 5). The second level of analyses (type, intensity, and timing) were assessed at both α=0.05 and α=0.005 based on a Bonferroni correction (Tables 4 and 5).
Table 5 Adjusted models for the relation between type, intensity, and life periods of physical activity and global cognition when also adjusting for respective types, intensities, or life periods
a
Multivariable adjusted for age at the time of LTPAQ interview, sex, NAART,
$\dot{\rm V}{\rm O}_{2} $
max, BMI, respective physical activity groupings, and interaction terms (age-sex, age-predictor, sex-predictor, age-sex-predictor).
b
Multivariable adjusted for age at the time of LTPAQ interview, sex, NAART,
$\dot{\rm V}{\rm O}_{2} $
max, BMI, past year PA, and interaction terms (age-sex, age-predictor, sex-predictor, age-sex-predictor).
*p<0.05.
δ*p<0.005 (Bonferonni corrected p-value).
MET(s)=metabolic equivalents.
Current Measures and Current Cognitive Function
In addition to long-term effects of PA, we also examined relatively acute effects of fitness and past year PA. All post hoc analyses were adjusted for sex, NAART, waist circumference, blood pressure, smoking status, and alcohol consumption, and revealed a statistically significant relation between past year PA and global cognition (p=.019) (Table 6), with no mediating effects of cerebrovascular indices (data not shown). Past year PA was associated with current
$\dot{\rm V}{\rm O}_{2} $
max (p<.0001). In contrast, relations were observed between current fitness (
$\dot{\rm V}{\rm O}_{2} $
max), global cognition, and cerebrovascular health using mediation analysis (Figure 2b). For every one unit increase (ml/kg/min) in
$\dot{\rm V}{\rm O}_{2} $
max, there was a 0.57 increase in global cognition Z-score (p<.001) (Table 6); for every one unit increase in
$\dot{\rm V}{\rm O}_{2} $
max, there was a 0.60 unit increase in CVC (p=.004) and a 0.0081 unit increase in
$\bar{\rm V}{\rm P}$
(p=.001) (Supplementary Table 2). Finally, with every one unit increase in CVC, there was a 13.31 increase in global cognition Z-score (p=.005) while for
$\bar{\rm V}{\rm P}$
there was a 0.14 increase in global cognition Z-score (p=.014) (Supplemental Table 3). The first three pathways of mediation were statistically significant; therefore, the estimated indirect effect of CVC was calculated and accounted for approximately 13.3% of the total effect of
$\dot{\rm V}{\rm O}_{2} $
max on cognition, while
$\bar{\rm V}{\rm P}$
accounted for 8.4%. However, the Sobel test revealed that the indirect effect of CVC and
$\bar{\rm V}{\rm P}$
on the relation between
$\dot{\rm V}{\rm O}_{2} $
max, and cognition was not statistically significant (p=.089 and p=.17, respectively).
Table 6 Adjusted overall model for associations between past year physical activity and global cognition and current fitness (
$\dot{\rm V}{\rm O}_{2} $
max) and global cognition
Note. Multivariable adjusted for sex, NAART, waist circumference, blood pressure, smoking status. and alcohol consumption. R2 attributable to Past year PA=0.15%, Adjusted model R2=21.4%. R2 attributable to
$\dot{\rm V}{\rm O}_{2} $
max=0.44%, Adjusted model R2=23.6%.
PA=physical activity; NAART=North American Adult Reading Test;
$\dot{\rm V}{\rm O}_{2} $
max=maximal aerobic capacity.
Discussion
The primary finding for this study demonstrated that greater total lifetime PA is associated with better global cognitive performance. The impact of lifetime PA on global cognitive performance differed for males and females as a result of an interaction between age, sex, and lifetime PA. In our sample, males followed the expected trajectory of increased global cognitive performance with increased levels of lifetime PA, whereas females did not (Figure 5). Many studies have reported no difference between males and females when assessing the relation between PA and cognition (Chang et al., Reference Chang, Jonsson, Snaedal, Bjornsson, Saczynski, Aspelund and Launer2010; Middleton, Mitnitski, Fallah, Kirkland, & Rockwood, Reference Middleton, Mitnitski, Fallah, Kirkland and Rockwood2008; Wendell et al., Reference Wendell, Gunstad, Waldstein, Wright, Ferrucci and Zonderman2014). Furthermore, evidence from all-male studies suggests a positive associations between increasing PA levels and decreased risk of cognitive impairment, Alzheimer’s disease, and related dementias (Abbott et al., Reference Abbott, White, Ross, Masaki, Curb and Petrovitch2004; van Gelder et al., Reference van Gelder, Tijhuis, Kalmijn, Giampaoli, Nissinen and Kromhout2004). Our findings are contrary to the hypothesis that the effects of PA on cognitive functioning may be more prominent in women (Brown et al., Reference Brown, Peiffer and Martins2013; Colcombe & Kramer, Reference Colcombe and Kramer2003; Ho et al., Reference Ho, Woo, Sham, Chan and Yu2001; Laurin et al., Reference Laurin, Verreault, Lindsay, MacPherson and Rockwood2001). It has been proposed that women should show greater cognitive change in response to exercise as compared to men, due to the variation in cognitive strengths between men and women. This differential sensitivity may be a result of organizational effects of sex hormones (Kimura, Reference Kimura2002; Kimura & Hampson, Reference Kimura and Hampson1994; Kramer & Erickson, Reference Kramer and Erickson2007). Declines in circulating sex steroid hormones associated with aging have been implicated as underlying age-related changes in brain health that differ for males and females. Specifically, there is evidence of altered CBF (Matteis et al., Reference Matteis, Troisi, Monaldo, Caltagirone and Silvestrini1998), volume of the frontal and temporal lobes of the brain (Cowell et al., Reference Cowell, Turetsky, Gur, Grossman, Shtasel and Gur1994), and increases in an indicator of cortical atrophy, increased ventricular and peripheral cerebrospinal fluid volume (Coffey et al., Reference Coffey, Lucke, Saxton, Ratcliff, Unitas, Billig and Bryan1998) in older men, compared to women.
Our finding that increased PA during lifetime is associated with greater cognitive functioning in middle-aged and older adults accords with findings from a recent systematic review and meta-analysis by Blondell et al. (Reference Blondell, Hammersley-Mather and Veerman2014). Previous literature indicates that increasing levels of PA in later life improved cognitive performance, decreased the risk for cognitive decline after the age of 55 years (Brown et al., Reference Brown, Peiffer and Martins2013; Buchman et al., Reference Buchman, Boyle, Yu, Shah, Wilson and Bennett2012; Laurin et al., Reference Laurin, Verreault, Lindsay, MacPherson and Rockwood2001; Middleton et al., Reference Middleton, Mitnitski, Fallah, Kirkland and Rockwood2008; van Gelder et al., Reference van Gelder, Tijhuis, Kalmijn, Giampaoli, Nissinen and Kromhout2004), and decreased the risk of developing dementia and Alzheimer’s disease (Abbott et al., Reference Abbott, White, Ross, Masaki, Curb and Petrovitch2004; Buchman et al., Reference Buchman, Boyle, Yu, Shah, Wilson and Bennett2012; de Bruijn et al., Reference de Bruijn, Schrijvers, de Groot, Witteman, Hofman, Franco and Ikram2013; Laurin et al., Reference Laurin, Verreault, Lindsay, MacPherson and Rockwood2001; Scarmeas et al., Reference Scarmeas, Luchsinger, Schupf, Brickman, Cosentino, Tang and Stern2009). To date, no study has sought to combine information on the duration, frequency, and intensity of all types of activity to create one comprehensive measure describing the average volume of PA completed across the lifetime. Therefore, using the measure of average MET-hr/week/year of PA provides novel insight into the positive effects that higher levels of PA throughout life will have on cognitive abilities in older age.
Timing and intensity of activity were also deemed to be important in the relation between PA and cognitive performance. When assessing the association between lifetime engagement in low, moderate, or vigorous PA and cognitive functioning, only vigorous intensity activity was associated with better global cognitive performance. The average hr/week/year of vigorous intensity PA over lifetime had the greatest impact on global cognition Z-scores, indicating vigorous intensity activity may be the most important for maintaining cognitive functioning into old age. Our findings align with the results of the Finnish Twin Cohort study (Iso-Markku, Waller, Kujala, & Kaprio, Reference Iso-Markku, Waller, Kujala and Kaprio2015) showing that participants who persistently engaged in vigorous activity in midlife had a decreased risk of mortality from dementia (Iso-Markku et al., Reference Iso-Markku, Waller, Kujala and Kaprio2015). Our findings also provide evidence that vigorous intensity activity above low or moderate intensity will help reduce cognitive deficit and potentially other devastating effects of dementia. Furthermore, lifetime recreational (including transportation) activity was independently associated with greater cognitive functioning. This finding aligns with previous investigations suggesting that current engagement in leisure time or recreational activities is associated with better cognitive function into older age (Rovio et al., Reference Rovio, Kareholt, Helkala, Viitanen, Winblad, Tuomilehto and Kivipelto2005; Weuve et al., Reference Weuve, Kang, Manson, Breteler, Ware and Grodstein2004; Yaffe et al., Reference Yaffe, Barnes, Nevitt, Lui and Covinsky2001). This finding promotes the etiologic role of sustained, planned PA throughout life for delaying or preventing age-related cognitive decline.
Physical activity in early childhood to midlife (age 0–35 years) had a substantial impact on better global cognitive performance at an older age. Several studies have assessed PA in midlife (i.e., 40–60 years) and have reported an association between greater midlife activity and a reduced incidence of dementia (Andel et al., Reference Andel, Crowe, Pedersen, Fratiglioni, Johansson and Gatz2008; Chang et al., Reference Chang, Jonsson, Snaedal, Bjornsson, Saczynski, Aspelund and Launer2010; Kareholt et al., Reference Kareholt, Lennartsson, Gatz and Parker2011; Rovio et al., Reference Rovio, Kareholt, Helkala, Viitanen, Winblad, Tuomilehto and Kivipelto2005; Sun et al., Reference Sun, Townsend, Okereke, Franco, Hu and Grodstein2010; Tolppanen et al., Reference Tolppanen, Solomon, Kulmala, Kareholt, Ngandu, Rusanen and Kivipelto2015). These results are contradictory with our findings of PA in early life being associated with improved cognitive performance in middle and older age. Relations may not have been observed in midlife as a result of our sample being sedentary before enrollment (age 50–65 years or midlife for many participants). It is likely that many individuals in our sample have been more sedentary in years leading up to the assessment than in early life. However, our findings closely align with Middleton (2010) who assessed self-reported PA at multiple points in time, and found that the strongest relationship was between the level of PA during adolescence and cognitive status during later life. Therefore, more research is warranted to determine the life periods in which PA levels contribute to better cognitive abilities.
Although lifetime PA contributes to better cognitive performance in middle and older age, no associations were observed with measures of cerebrovascular function. Based on previous research, we hypothesized that measures of acute fitness or high levels of current PA may have more impact on cerebrovascular health. Post hoc analyses were performed assessing past year PA and current fitness, revealing
$\dot{\rm V}{\rm O}_{2} $
max measured at baseline (i.e., in later life) was more predictive of cerebrovascular health than lifetime or past-year measures of PA. Similarly, Bailey et al. (Reference Bailey, Marley, Brugniaux, Hodson, New, Ogoh and Ainslie2013) reported that adults who were more active doing recreational activities over lifetime (confirmed with
$\dot{\rm V}{\rm O}_{2} $
max) had better cerebrovascular function. Previous literature has linked current fitness levels to enhanced cerebrovascular function (Ainslie et al., Reference Ainslie, Cotter, George, Lucas, Murrell, Shave and Atkinson2008; Barnes et al., Reference Barnes, Taylor, Kluck, Johnson and Joyner2013; Burdette et al., Reference Burdette, Laurienti, Espeland, Morgan, Telesford, Vechlekar and Rejeski2010) and improved cognition (Colcombe et al., Reference Colcombe, Kramer, Erickson, Scalf, McAuley, Cohen and Elavsky2004; Forbes et al., Reference Forbes, Thiessen, Blake, Forbes and Forbes2013; Kramer et al., Reference Kramer, Hahn, Cohen, Banich, McAuley, Harrison and Colcombe1999; Prakash et al., Reference Prakash, Voss, Erickson and Kramer2015, Reference Prakash, Voss, Erickson, Lewis, Chaddock, Malkowski and Kramer2011). Therefore, by determining that measures of cerebrovascular health (
$\bar{\rm V}{\rm P}$
and CVC) partially mediate the relation between
$\dot{\rm V}{\rm O}_{2} $
max and global cognitive performance provides evidence that cerebrovascular health may play a part in the relation between current fitness and cognition.
A major strength of this study was the comprehensive, reliable measure of lifetime total PA (Friedenreich et al., Reference Friedenreich, Courneya and Bryant1998), which captured types, intensities, and time periods in life when PA could be most beneficial for reducing cognitive decline associated with aging. Additionally, we were not limited in our analyses to fully explore the association between PA, cerebrovascular measures, and cognition. An additional strength was the unique testing protocols, such as the extensive neuropsychological test battery for cognitive abilities and the CBF test for cerebrovascular measurements (Brown et al., Reference Brown, McMorris, Longman, Leigh, Hill, Friedenreich and Poulin2010; Tyndall et al., Reference Tyndall, Davenport, Wilson, Burek, Arsenault-Lapierre, Haley and Poulin2013). Finally, the quality and extent of participant information collected permitted a full assessment of covariates in the analysis.
Lifetime PA and other factors (i.e., age, sex, NAART,
$\dot{\rm V}{\rm O}_{2} $
max, BMI, age-sex, age-lifetime PA, sex-lifetime PA, and age-sex-lifetime PA interactions) explain between 34.0% and 36.1% of the variance in global cognition. A better model fit could be potentially obtained by considering factors such as current social activities, history of depression, or family history. Additionally, it is possible that other factors may have mediated the relation between lifetime PA and cognitive functioning. In the future when assessing lifetime measures of PA, it would be beneficial to explore biological and environmental factors as mediators that may have an impact over one’s lifetime such as history of socio-economic status (impacting opportunities over one’s lifetime) or intellectual ability in the individual’s youth (impacting decisions to participate in PA).
We did not collect information on hormone levels at baseline; therefore, we are limited in our ability to fully explore the reason for observed sex differences between males and females. Further our findings have limited generalizability to all older adults given that it is based on a highly educated, mostly Caucasian volunteer sample. There is the possibility of misclassification of lifetime PA since participants may not have been able to recall their activities accurately. However, given that questionnaire has high reproducibility (Friedenreich et al., Reference Friedenreich, Courneya and Bryant1998), we anticipate the effect of this measurement error to be minimal and any misclassification to be non-differential with the effect of reducing our ability to show a relation between PA and cognition suggesting that these results are conservative estimates of the true effect. Additionally, the questionnaire has not been tested for validity. An ideal but unrealistic validation study would require historical PA data on a cohort of study participants assessed repeatedly over their lifetimes to capture the same data as are assessed in the LTPAQ used here. It is possible that PA from earlier in life is not accurately captured.
We used the Bonferroni correction for multiple regression comparisons to reduce the risk of type I error. Using this correction resulted in loss of significance for some findings. Although we reduced type I error with a conservative α level, these corrections reduce statistical power and increase the risk of type II error; therefore, the corrected α level should be interpreted with caution (Rothman, Reference Rothman1990). Finally, a cross-sectional study design cannot assess causality. Recent research by Belsky et al. (Reference Belsky, Caspi, Israel, Blumenthal, Poulton and Moffitt2015) demonstrated that children with higher levels of cognitive ability choose healthier lifestyles, and also had improved cognitive functioning later in life. These findings suggest that causality cannot be concluded. However, there has also been evidence suggesting cognitive abilities are not consistent over one’s lifespan and speculate that both innate abilities and lifestyle must be considered together rather than looking at them as separate entities for determining cognitive changes with aging (Deary, Whiteman, Starr, Whalley, & Fox, Reference Deary, Whiteman, Starr, Whalley and Fox2004). A prospective longitudinal study with multiple data collection and testing periods would be required to address this question more fully.
This study adds to previous evidence that PA protects against poor cognitive functioning in older age while providing new insight into the importance of lifetime PA and the potential mediating effects of cerebrovascular health. We found that total lifetime PA, specifically participation in recreational activities, hours spent in vigorous intensity activity over lifetime, and PA done in early childhood to midlife, are associated with improved global cognitive performance in older age. Although cerebrovascular health did not mediated the association between lifetime or past year PA and global cognition, it appeared to partially mediate the association between current fitness levels and global cognition in middle-aged and older adults. Thus, higher levels of PA throughout life and in the past year of life are associated with better cognitive functioning in older age, while current fitness levels may be more important for improved cerebrovascular health in addition to cognitive functioning. This difference emphasizes the finding that cerebrovascular health may be more closely linked to physiological measures of fitness than subjective measures of activity.
Acknowlegments
The authors thank all collaborators, staff, and participants of the Brain in Motion project, and staff from the department of Cancer Epidemiology and Prevention Research at Alberta Health Services for assistance with data management and analysis. The authors report no conflicts of interest. Author contributions statements: C.M.F., G.A.E., R.S.L., D.B.H., M.D.H., T.J.A., and M.J.P. conceived the experimental design. M.J.P. had full access to all of the data in the study and takes responsibility for the integrity of the data and accuracy of the data analysis. M.J.P. and C.M.F. led, guided, and oversaw the analytical (data collection, validation, final analyses) work and manuscript writing, all of which was conducted by S.J.G. T.T.S. guided and oversaw the statistical analysis. A.V.T., M.H.D., L.L.D. assisted with data collection. B.J.W. provided medical coverage for fitness testing (
$\dot{\rm V}{\rm O}_{2} $
max). Primary supervision for S.J.G. was provided by M.J.P. and C.M.F. S.J.G. wrote the first draft of the manuscript, and C.M.F., T.T.S., R.S.L., L.L.D., M.H.D., A.V.T., G.A.E., D.B.H., M.D.H., J.S.P., B.J.W., and M.J.P. reviewed and edited the manuscript. Funding Sources: SJG is supported by the Brenda Strafford Foundation Chair in Alzheimer Research (BSFCAR). CMF is supported by an Alberta Innovates-Health Solutions Health Senior Scholar Award and the Alberta Cancer Foundation’s Weekend to End Women’s Cancers Breast Cancer Chair. M.H.D. was supported by a Heart and Stroke Foundation of Canada (HSFC) (Grant number 13-0001867) and Canadian Institutes of Health Research (CIHR) Focus-on-Stroke Postdoctoral Fellowship. A.V.T. is supported by an Alzheimer Society of Canada Doctoral Award. T.T.S. received a University of Calgary Seed Grant Award. M.J.P. holds the BSFCAR. The funding for the study and all biochemical analyses was provided by CIHR (Principal Investigator=M.J.P.; Co-Applicants=C.M.F., G.A.E., D.B.H., and M.D.H.; Collaborators=T.J.A. and T.T.S.) and the BSFCAR. The funders played no role in the concept and design of this study, analysis or interpretation of the data, or drafting and critical revision of the manuscript.
Supplementary Material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S1355617715000880
