Enhanced task-related brain activation and resting perfusion in healthy older adults after chronic blueberry supplementation

Twenty-six participants (Table 1) completed this double-blind randomized controlled trial, which was approved by the Sport and Health Sciences Ethics Committee at the University of Exeter and performed in accordance with the Declaration of Helsinki. At an initial visit, after providing their written informed consent, participants completed the Addenbrooke’s Cognitive Examination III questionnaire (ACE-III, version A), and magnetic resonance imaging (MRI) safety screening questionnaire to ensure their suitability for the trial prior to completing the cognitive function tasks to familiarize themselves with the test procedures. The ACE-III questionnaire, developed at Neuroscience Australia, assesses 5 cognitive domains: attention, memory, verbal fluency, language, and visuospatial abilities. The test took approximately 15 min to administer, and scoring was performed after the visit following the validated protocol (Hsieh et al. 2013). A cut-off score of 88/100 was adopted to indicate cognitive impairment as recommended for a research context requiring high sensitivity (1.00) and lower specificity (0.96) (Hsieh et al. 2013). Exclusion criteria were cognitive impairment, any contraindications to MRI, consuming more than 5 portions of fruit per day, and an age of less than 65 years. No participants were excluded on the basis of their ACE-III score.

Participants were pair-matched for ACE-III score and randomised to a group, with participants and investigators blind to treatment. Measurements were taken before and after 12 weeks of supplementation with either BlueberryActive (CherryActive Ltd., Sunbury, UK) or an isoenergetic placebo. Participants were instructed to consume their habitual diet throughout.

Thirty millilitres of blueberry concentrate were consumed once per day for 12 weeks and provided 387 mg anthocyanidins (34 mg malvidin, 108 mg cyanidin, 41 mg pelargonidin, 63 mg peonidin, 86 mg delphinidin, and 55 mg petunidin; analysed by high performance liquid chromatography; Chandra et al. 2001) and 25.5 g carbohydrate. The placebo was a synthetic blackcurrant and apple cordial (Robinsons cordial, Britvic Ltd., Hemel Hempstead, UK) with sugar added to match blueberry energy content. Participants and researchers were unaware of the condition to which they were randomized, and blinding was maintained since participants were informed that the aim of the study was simply to test the effects of fruit concentrates on brain function.

Participants completed a battery of cognitive function tests (CogState Ltd., New Haven, Conn., USA) in a quiet room, after which a venous blood sample was taken prior to completing MRI testing. Participants were subsequently given 12 × 210 mL bottles of the appropriate liquid supplement that were identical in appearance for placebo and blueberry conditions and instructed to take a 30 mL dose diluted to 240 mL total volume with tap water every morning for the following 12 weeks. A measuring cup was provided for this purpose. Participants returned 12 weeks later to repeat all measurements, and they returned the bottles to allow assessment of compliance with the supplementation regime.

The battery of tests (CogState Ltd.) selected for the present study took approximately 35 min to complete and assessed a range of cognitive domains that have previously been shown to be affected by polyphenol supplementation: psychomotor function, visual processing, executive function, verbal and spatial memory, and working memory (for review see Lamport et al. 2014). Specifically the tests comprised a detection task (has a card turned over, reaction time) to assess psychomotor function; the Groton maze timed chase test (chasing a visual target, moves per second) to assess speed of visual processing; the Groton maze learning test with a delayed recall component (find a hidden pathway through the maze and then recall after a delay, number of errors and duration) to assess executive function and delayed recall; identification task (is a card red, reaction time) to assess attention; international shopping list task with delayed recall (learn and recall shopping list items, number of correct responses) to assess verbal learning and delayed recall; and 1-back and 2-back memory tasks (is the card the same as 1 or 2 back, reaction time and correct responses proportion) to assess working memory. The speed and accuracy of responses were quantified.

The cognitive task undertaken while acquiring functional MRI (fMRI) data was a numerical Stroop test (Kaufmann et al. 2008), chosen to challenge attention and memory. The task was selected based upon its independence on knowledge, training, practice, or educational level, particularly in terms of language or mathematical ability. In addition, the task only required the pressing of a single button of a pair of MRI-compatible response boxes (Cedrus, San Pedro, Calif., USA); one button was held in each hand to ensure ease of response within the scanner environment. The Stroop task involved the presentation of a series of single-digit number pairs shown side by side with participants requested to press the button held in the same hand as the side of the screen the larger number of the pair appeared. The definition of ‘larger’ as applied to the number pairs varied during the course of the experiment, with classification based upon either numerical value or physical size, and the experiment was split up into blocks with each block beginning with the presentation of a classification guide: either the word ‘Numerical’ or ‘Physical’ being presented indicating which parameter should be considered when deciding which number was ‘bigger’. The test was run using E-Prime version 2 software (Psychology Software Tools Inc., Sharpsburg, Penn., USA) with participant responses recorded allowing an assessment of accuracy and reaction times. In all cases participants practiced the task once they had been positioned within the scanner to ensure full familiarity with the test, the environment, and the method of responding. Responses were monitored and the test proper not begun until participants were consistently responding correctly.

For protein carbonylation detection, serum proteins were denatured using 12% sodium dodecyl sulfate and derivatised using 1X 2,4-dinitrophenylhydrazine solution at room temperature for 15 min and neutralised. Carbonylated proteins were subsequently detected using Western blot analysis (antidinitrophenylhydrazine: 1/2000 and anti, glyceraldehyde 3-phosphate dehydrogenase: 1/1000, respectively) as described in Hull et al. (2015); 4-hydroxynonenal was detected using anti-hydroxynonenal (1/500) and secondary anti-rabbit (1/2000). Serum glutathione was assayed using a commercially available reduced glutathione/glutathione disulfide ratio detection assay kit (Abcam, Cambridge, Mass, USA). Serum samples were also analysed for brain-derived neurotrophic factor (BDNF) by commercially available enzyme-linked immunosorbent assay (Chemikine, Darmstadt, Germany) and for C-reactive protein (CRP) using a turbidometric assay.

All MRI was undertaken in a 1.5 T Philips Gyroscan scanner with an 8-element head coil. Within the head coil a mirror assembly was mounted such that with their head within the coil participants were able to see a screen at the end of the scanner bed onto which visual-based cognitive tasks could be projected. The assessment of brain activity during a cognitive task was undertaken using a standard single shot echo-planar dynamic imaging sequence (repetition time = 3 s, echo time = 45 ms, resolution 2.5 mm × 2.5 mm × 3.5 mm, 39 contiguous transverse-oblique slices, field of view 230 mm × 230 mm, 64 × 64 within-plane matrix, 220 dynamics) while undertaking a numerical Stroop test (Kaufmann et al. 2008) (144 trials) where image presentation was alternated with the presentation of a fixation cross. This was followed by the acquisition of a high-resolution whole-brain T1-weighted anatomical scan (resolution 0.9 mm × 0.9 mm × 0.9 mm) and a pseudo-continuous arterial spin labelling (ASL) sequence to assess any modifications in brain perfusion at rest, consisted of labelling tag of 1800 ms applied to the carotid artery 20 mm inferior to a stack of 14 acquisition slices covering the whole brain. Data were recorded for 5 different delay times (400, 800, 1200, 1600, 2000 ms) with 30 pairs of control-tag images acquired for each delay time.

The assessment of brain activity during the cognitive task was undertaken by monitoring alterations in local blood delivery characteristics linked to task performance relative to that associated with the fixation cross by examining changes in image signal intensity. Analysis was undertaken using SPM8 software (The Wellcome Department of Cognitive Neurology, University College, London) a suite of MATLAB functions and subroutines. Preprocessing included slice time correction, spatial processing to correct for head movement and size, and warping to the Montreal Neurological Institute template (MNI305). Images were then convolved with a 3D Gaussian filter with an 8 mm full-width-at-half-maximum. The fMRI data were analyzed based on mass univariate (voxel-by-voxel) testing within the general linear model framework over the whole brain, treating each participant separately and constructing individual maps comparing the differences in response between the 2 visits. Statistical testing took place examining the differences in group responses for overall activation, size of hemodynamic response, and timing of hemodynamic activation with significant brain activation defined as arising within a region where the differences between groups in signal intensity or the time to signal peak gave rise to a p value <0.001 after no corrections had been made for multiple comparisons and the cluster size of the activated region was equal or greater than 10 voxels. For ASL analysis, ASL images were initially registered to the structural images within the FSL software package (FMRIB Software Library v5.0, Oxford University, UK) after which the structural images were registered into MNI space and the generated warping matrix applied to the ASL data so it was also present within MNI space. Difference images were obtained from the control-tag pairs followed by averaging over the 30 acquisitions at each delay time, such that a single difference image was obtained for each delay time. Quantitative perfusion calculations were then undertaken using the BASIL toolbox (Bayesian inference for arterial spin labelling MRI) (Chappell et al. 2009, 2010) within FSL. Regions of interest corresponding to the grey matter of the parietal, frontal, and occipital lobes were defined based upon the MNI structural atlas and the average perfusion values within each region determined.

Sample size calculations were based on change in region of interest-based fMRI signal with a population SD (<0.5%) (Zandbelt et al. 2008), although accurate calculations were hampered by the limited available data on the effects of blueberry polyphenols on fMRI. A sample size of 12 participants per group was estimated based on a moderate effect size of 0.5 with 80% power and α level of 0.05. Data are reported as mean ± SEM for baseline and postsupplementation. Cognitive function, cerebral perfusion (ASL), as well as serum BDNF and CRP concentration data were analysed by 2-way mixed model ANOVA (treatment vs time) to determine whether there were any statistically significant effects of time or treatment. The changes in cognitive performance and fMRI data from pre- to postsupplementation were compared across groups using 2-tailed independent sample t tests with equal variance.

There were no significant differences between groups for any of the baseline characteristics (Table 1).

No significant performance differences were seen between pre- versus postsupplementation visits or between groups for the number of correct responses while undertaking the numerical Stroop test. Over all cases, group accuracy percentages ranged between 98.1% and 98.8%. However, significant increases in brain activation responses were found in a number of task-associated regions following blueberry supplementation compared with placebo relative to the baseline visits (Brodman areas 4, 6, 10, 21, 40, 44, 45, precuneus, anterior cingulate, insula and thalamus, all p < 0.001, Fig. 1). In contrast, no significant increases in brain activity were observed following placebo compared with blueberry supplementation relative to baseline.

Resting-state quantitative perfusion of the gray matter increased significantly after blueberry supplementation but not placebo supplementation in the parietal (p = 0.013, Fig. 2b) and occipital lobes (p = 0.031, Fig. 2c). There was no change in resting-state perfusion in the frontal lobes for either condition (Fig. 2a).

There was a significant decrease in serum glutathione concentration in both conditions (placebo: from 73.7 ± 3.6 to 64.4 ± 2.6; blueberry: from 70.0 ± 4.0 to 65.1 ± 3.6 μmol·L⁻¹; main time effect p < 0.001), which tended to be smaller in the blueberry condition (placebo: –11.7 ± 2.8%; blueberry: –6.5 ± 2.4%; p = 0.09). However, an overall change in protein carbonylation HNE adduct or malonaldehyde formation between these groups was not evident (data not shown). There was no significant change over time in either serum hsCRP (placebo: 1.7 ± 0.6 to 1.4 ± 0.4; blueberry: 1.4 ± 0.5 to 1.4 ± 0.4 mg·L⁻¹) or BDNF (placebo: 103.7 ± 10.1 to 98.8 ± 9.0; blueberry: 88.6 ± 7.3 to 97.4 ± 9.6 ng·mL⁻¹) concentration, nor was there any difference between conditions.

Performance of the Groton maze learning task (accuracy, p = 0.005), international shopping list task (p = 0.002), and international shopping list with delayed recall (p = 0.004, Table 2) improved over time, but there was no significant difference in this improvement among groups. Performance of the 1-back test tended to improve to a greater extent in the blueberry group but this was not statistically significant (speed, p = 0.094).

The percentage change in performance of the 2-back tests showed weak evidence for improvement in the blueberry versus placebo groups (reaction time: placebo: 0.4 ± 0.4% vs blueberry: –1.0 ± 0.7%; p = 0.09; accuracy: placebo: –3.8 ± 2.5%; blueberry: 3.6 ± 2.7%; group by time interaction effect: p = 0.05). The change in performance for the other cognitive function tests were not significantly different between placebo or blueberry supplementation.

The authors report no conflicts of interest associated with this manuscript.