Apparently healthy young adults (aged 18–35 years) with a body mass index (BMI) between 18.5 and 30.0 kg/m2 were recruited through local advertisements and social media. Participants were invited for a screening visit to evaluate their eligibility and to ensure proper familiarization with the cognitive performance measurements. Exclusion criteria were pre-existing neurological illnesses, use of medication or dietary supplements that might influence endpoints of the study (such as antidepressants or blood pressure medication), high blood pressure (> 140 mmHg systolic and/or > 90 mmHg diastolic blood pressure), pregnancy, smoking, and abuse of alcohol (> 20 alcoholic units/week) or drugs. All participants gave written informed consent before data collection. The study was approved by the Medical Ethics Committee of University Hospital Maastricht and Maastricht University (METC azM / UM) and performed at the University of Maastricht between October 2020 and April 2021 in accordance with the 1964 Declaration of Helsinki and its later amendments. The study was registered online at ClinicalTrials.gov as NCT03793777.

A randomized, double-blind, placebo-controlled, cross-over study was performed. After successful screening, each participant was randomly allocated to receive 1 week of either AME or placebo, separated by a wash-out period of at least 2 weeks. At baseline and after 1 week of intervention, participants visited the research facilities. Anthropometric parameters, cognitive performance, and vascular function measurements were performed in temperature-controlled rooms of the Metabolic Research Unit Maastricht (MRUM), and fasted blood samples were collected. All study visits were performed in the morning in a fasted state, and different study visits for each participant occurred at the same time of the day. An electronic data capture system (Castor EDC, Amsterdam, the Netherlands) was used for data collection and the study was monitored by the Clinical Trial Center Maastricht (CTCM). The randomization schedule was computer-generated and was performed by an independent researcher, using random and concealed block sizes, and stratifying for sex.

During the study, participants were provided with a list of foods containing anthocyanins and were instructed to refrain from these products during the study period. Furthermore, participants were instructed to avoid strenuous physical activity and alcohol on the 2 days preceding each test day, and to arrive fasted in the morning of the measurements.

The study product was an AME containing 24% anthocyanins (14.4% cyanidin-3-galactoside and 9.6% of other cyanidin-3-glycosides, [i.e. cyanidin-3-arabinoside, cyanidin-3-xyloside, cyanidin-3-glucoside]) which was provided by BioActor BV (Brainberry®; Maastricht, The Netherlands). A daily dose of 750 mg AME was provided in three capsules containing in total 180 mg anthocyanins per day. This dose was based on previous acute and short-term research with anthocyanin interventions in young populations showing improvements in cognitive performance [14,15,16]. Identical numbers of cellulose-containing capsules were used as placebo. AME and placebo capsules were opaque and uniform in appearance. The capsule jars each contained 30 capsules and were blinded, displaying only the participant number and the intervention period on the label. Participants were instructed to consume three capsules daily before breakfast, with 200 mL water, and to note daily intake and any deviations in a supplementation logbook. Remaining capsules at the end of the intervention period were returned to the study facility to monitor compliance, which was considered valid if > 85%.

Cognitive performance was assessed using the Cambridge neuropsychological test automated battery (CANTAB) [17]. This is a validated and computerized assessment that we have previously used to assess three main cognitive domains: attention and psychomotor speed, memory, and executive function [18]. The motor screening test (MOT) was performed as the first test in every session and was used to get familiar with the CANTAB system. Attention and psychomotor speed were assessed using the five-choice reaction time (RTI) test. In the RTI test, the time from stimulus to release of the response button (reaction time) and the time from release of the response button to selection of target (movement time) of correctly assessed trials is determined. Measurements in the memory domain were the delayed matching to sample (DMS) and paired associates learning (PAL) test. For the DMS, the combined percentage of correct trials after 0 s, 4 s, and 12 s of delay were recorded. The PAL included a first-attempt memory score and total errors on the 12-item trial. Executive function was measured with the multitasking test (MTT) and spatial working memory test (SWM). The MTT provided the mean response latency, incongruency cost, multitasking cost, and total errors. For the SWM, total errors on the 12-item trial and strategy score were included.

Mood was assessed using 100 mm visual analogue scales, including eight mood states: afraid, angry, confused, energetic, happy, sad, tense, and tired [19]. The cognitive failures (CFQ) was used to assess subjective cognitive failure in daily life [20].

Measurements were performed in a supine position after an acclimatization period of at least ten minutes. First, office blood pressure was determined in fourfold, of which the first measurement was disregarded (Omron 705IT, Hoofddorp, The Netherlands). Radial artery pulse wave analyses (PWA) of the brachial artery were performed in triplicate to determine mean arterial pressure (MAP), near the elbow and wrist of the arm using a tonometer (SphygmoCor v9, AtCor Medical, West Ryde, Australia). Consequently, the central augmentation index corrected for heart rate (cAIxHR75) was calculated using the difference between the first and second peaks of the central arterial waveform. Furthermore, carotid-to-femoral pulse wave velocity (cfPWV) was measured in triplicate as a measure of arterial stiffness, as described previously [21].

Microvascular retinal calibers were determined using a fundus camera (Topcon TRC-NW-300, TopCon Co., Tokyo, Japan), Vascular images of the optic disc were taken and images from both study periods were analyzed simultaneously using Interactive Vessel Analyzer software (IVAN, University of Wisconsin, Wisconsin, USA) to ensure that the selected segments were identical in all images of a participant. Using the Parr-Hubbard formula [22], the mean central retinal arteriolar and venular equivalents (CRAE and CRVE, respectively) and the arteriolar-to-venular ratio (AVR) were calculated.

Fasting blood samples were collected in serum separator tubes (BD Vacutainer, NJ, USA). Serum tubes were centrifuged after exactly one hour of coagulation (room temperature, 1300xg, 10 min) as clotting time could affect BDNF concentrations in serum [23]. Serum aliquots were stored for further analysis at the end of the trial. BDNF concentrations were determined using an enzyme-linked immunosorbent assay (Duoset ELISA, R&D systems, Minneapolis, MN, USA), according to the manufacturer’s protocol.

An expected effect size of 0.602 was determined, based on the change from baseline on repetition seven of the digit vigilance test (a measure of cognitive function, which was defined as the primary study outcome) in a study by Watson et al. [24]. Using a power of 90%, and a two-sided alpha of 0.05, a sample size of at least 31 participants was required. To account for expected drop-outs, a total of 36 participants were included. Statistical analyses were performed using IBM SPSS Statistics (26.0, IBM Corporation, Armonk, NY, USA).

Statistical analyses were performed using linear mixed models including intervention, period, and sex as fixed factors, baseline values as covariate, and participant and intercept as random factors. Raw scores of the post-intervention measures were used with the corresponding baseline values as covariates. The three-way (intervention*period*sex) and two-way interactions (intervention*period, intervention*sex, and period*sex) were omitted from the model if not significant following a top-down approach. Pearson correlations were determined between significant changes observed in cognitive performance and those in serum BDNF concentrations. Data were reported as unadjusted means ± SDs, unless stated otherwise. For all analyses, two-sided p values < 0.05 were considered statistically significant.