Anthocyanins, a subclass of the flavonoids, are found in foods such as berries and fruits and information regarding their content in food can be found in an online phenol-explorer (Neveu et al., 2010). Anthocyanins have been shown in previous studies to have a number of positive health effects, such as improving the blood lipid profile (Qin et al., 2009; Li et al., 2015), and also fasting serum glucose and glycosylated hemoglobin (HbA1c) in diabetic patients (Li et al., 2015; Yang et al., 2017), and have anti-inflammatory effects (Xia et al., 2007; Spencer et al., 2012; Spagnuolo et al., 2017). Anthocyanins can improve endothelial and vascular function (Rodriguez-Mateos et al., 2013, 2016, 2019), and can cross the blood–brain barrier (Faria et al., 2014) and thus may reduce neurodegenerative and cerebrovascular changes and possibly protect against cognitive decline and dementia. Interestingly, some studies have found that food-based anthocyanins can improve memory functioning in older adults with mild cognitive impairment (MCI; Krikorian et al., 2010a,b; Hein et al., 2019). However, these studies were based on relatively small samples, and had a short duration. In addition, food-based anthocyanin supplementation leads to heterogeneity, i.e., variations in types of food sources, concentration, and dose of anthocyanins.

In this exploratory open-label pilot study, we aimed as a primary endpoint to examine potential changes in dementia-relevant mechanisms after 16 weeks of treatment with purified anthocyanin containing capsules, in people with increased risk of dementia. As a secondary endpoint we also explored the potential change in a battery of cognitive tests.

During the period May 2015 to September 2016, 33 participants started anthocyanin supplementation, of whom 27 (8 MCI and 19 CAD) completed the study (6 were excluded for administrative and logistical reasons) (Figure 1). The NC were included in the period December 2016 to February 2017. No subjects with mild dementia were included.

Baseline characteristics are shown in Table 1.

Only IL-8, MCP-1, CCL-5/RANTES [regulated on activation, normal T-cell expressed and secreted (RANTES)] and TNF were available for statistical analyses, as the other inflammation markers did not reach measurement thresholds. The findings are summarized in Table 2 and Supplementary Figures S1–S3.

The only significant between-group difference was for difference were for ΔMCP-1 (difference from baseline to study end) (p = 0.011) and Δfasting glucose (p = 0.003) (Table 2).

When analyzing the groups separately, significant increases were found in total cholesterol and triglycerides in the anthocyanin supplementation group (AG) from baseline to study end (Table 2), and MCP-1 which increased in the NC group (Table 2).

No significant changes were found for fasting glucose and HbA1c in the AG group.

A total of 29 plasma anthocyanin metabolites were quantified (Table 3 and Supplementary Figure S4). When comparing the two groups, a statistically significant difference was found for two metabolites (o-Coumaric acid and Dihydroferulic acid-4-O-Sulfate), which both had a larger decrease in the AG than in the NC group (Table 3).

In the AG, there was a statistically significant increase in five of the metabolites (Pyrogallol-2-O-sulfate, Protocatechuic acid-3-O-sulfate, Pyrogallol-1-O-sulfate, Ferulic acid-4-O-β-D-glucuronide, Isoferulic_acid-3-O-β-D-glucuronide) and a statistically significant decrease in five other metabolites (3-Hydroxyhippuric acid, 4-Hydroxybenzaldehyde, Dihydroferulic acid-4-O-Sulfate, m-Coumaric acid, o-Coumaric acid) after 16 weeks of anthocyanin consumption in comparison with baseline.

In the NC group, there was a statistically significant decrease in four metabolites (Protocatechuic acid, 3,4-Dihydroxyphenylacetic acid, 4-Hydroxybenzaldehyde and o-Coumaric acid), whereas there were no statistically significant increases in any of the metabolites.

The cognitive test scores improved in the intervention group, with improvements for CERAD learning (p = 0.016), recall (p < 0.001) and recognition (p = 0.047) and for STROOP test word (p < 0.001) and color (p = 0.044) (Table 4 and Supplementary Figures S5–S7).

Overall, findings using parametric analyses differed only marginally from the non-parametric findings reported above (Supplementary Tables S1, S2).

The compliance was good. More than 85% of the participants returned at least 90% of the empty blister packages. The anthocyanins were well tolerated, and none of the participants withdrew due to adverse effects. Blood tests taken for safety reasons were all within a clinically acceptable range. Increased bleeding tendency was not observed.

In this pilot study anthocyanin supplementation was well tolerated, without any AE, and the compliance was good. This indicates that larger RCTs might be feasible, to confirm exploratory results in the current pilot study.

Our findings are somewhat inconclusive. While some cognitive improvements were observed in the AG, there were no significant changes in serum levels of some risk factors for dementia; i.e., fasting glucose, HbA1c or pro-inflammatory cytokines. There was a non-significant increase in serum levels of MCP-1 in the AG and a significant increase in the NC during the study period. The between-group difference in Δ serum levels of MCP-1 was statistically significant.

Furthermore, we observed a significant increase in serum levels of total cholesterol and triglycerides in AG. The lipid profile of the NC group did not change significantly, and since we have no information about statin use or use of other lipid lowering medications in the NC group, the observed difference should be interpreted cautiously.

Previous studies using Medox^® have shown a statistically significant increase in HDL-cholesterol (Qin et al., 2009; Zhu et al., 2011; Hassellund et al., 2013) and a decrease in LDL-cholesterol (Qin et al., 2009; Zhu et al., 2011). This is of clinical interest, as higher HDL-cholesterol is associated with lower cardiovascular risk (Barter et al., 2007), while high levels of total cholesterol, triglycerides and LDL are associated with higher cardiovascular risk (Stone et al., 2014).

The differences in the findings between our study and these previous studies might be due to differences in participants, as well as in anthocyanin supplementation dose and duration, or other factors. Furthermore, other studies included dyslipidemic and hypercholesterolemic participants not using statins or any other lipid lowering treatment (Qin et al., 2009; Zhu et al., 2011), whereas in our study, the median cholesterol at baseline in the intervention group was 4.0 mmol/l, in addition 70% were taking statins or other lipid lowering medication.

Regarding the inflammation markers, RANTES promotes activation and migration of leukocytes and mediates neuroinflammation and brain microvascular dysfunction (Appay and Rowland-Jones, 2001; Dénes et al., 2010; Yilmaz and Granger, 2010). As there was a significant between-group difference for ΔMCP-1, our results are partly consistent with findings in a randomized, double-blind trial in hypercholesterolemic individuals consuming purified anthocyanins for 24 weeks (Song et al., 2014), and in a parallel-designed, placebo-controlled trial (Karlsen et al., 2007). Other studies did not report a reduction of pro-inflammatory mediators after anthocyanin supplementation (Hassellund et al., 2013; Kent et al., 2015). Therefore, the anti-inflammatory effect of anthocyanins and the potential to reduce neuroinflammation and brain microvascular dysfunction associated with cognitive decline in adults at risk of dementia (Grammas, 2011) should be studied in larger randomized studies.

The beneficial effect of anthocyanins might possibly be due to their degradation products and metabolites (Feliciano et al., 2016) as absorption of intact anthocyanins is reported to be low (Zhong et al., 2017). Rodriguez-Mateos et al. quantified metabolites in plasma after blueberry consumption, and showed that anthocyanin-derived metabolites correlated with in vivo effects (Rodriguez-Mateos et al., 2013, 2019). Furthermore, circulating anthocyanin metabolites were shown to improve vascular function when injected in an in vivo model of vascular function, and cardiovascular benefits after consumption of anthocyanins were linked with anthocyanin metabolites as mediators of change in cellular gene programs (Rodriguez-Mateos et al., 2019).

In our study, we were able to measure a total of 29 anthocyanin metabolites. The results were conflicting, as we found various anthocyanin metabolites to be both significantly increasing and decreasing in the AG. In the NC group as well, we found significant changes. This is in line with findings in another study reporting an increase of some metabolites, and a decrease in others after ingestion of anthocyanins over time (Feliciano et al., 2016). A possible explanation for our results could be high inter-individual variability and the influence of the background diet on the concentration of these compounds in blood, as some of these metabolites could arise from the consumption of other anthocyanin-rich foods such as berries and red wine or other food components in the diet. It is also possible that the results could be related to the handling and analysis of the blood samples. As far as we know, the presence of anthocyanin metabolites in plasma after Medox^® use is reported in only one other clinical study, where 8 out of 17 anthocyanin metabolites were found to be significantly increased. However, this was analyzed in blood samples collected 1–3 h after ingestion of the morning dose of anthocyanins (320 mg), and not after daily consumption and an overnight fast (Hassellund et al., 2013).

The improvement on several cognitive tests should be interpreted cautiously due to potential learning effects related to the relatively short test–retest interval, and the lack of a comparison group. Nonetheless, our results are in line with previous smaller studies involving participants with MCI, reporting improved cognition after ingestion of anthocyanins (Krikorian et al., 2010a,b). Our results are also in line with the results of a randomized, double-blinded, placebo-controlled study reporting improved episodic memory after 3 months of anthocyanin supplementation (Whyte et al., 2018).

Intake of anthocyanin capsules in the dosage of 320 mg/day appears to be well tolerated and safe. None of the safety blood tests were found to be out of a clinically acceptable range or necessitating medical follow-up after study-end. This is consistent with previous studies (Karlsen et al., 2007; Qin et al., 2009).

The major limitations of this pilot study are the non-randomized open-label design, the small sample size, and the relatively short intervention period. Although all the participants were told not to change their lifestyle during the intervention period, we have no data on this. Further we had no detailed dietary assessment, and thus we cannot exclude the possibility of differences in the background diet between the groups before or during the study. However, the participants were instructed to maintain their dietary and lifestyle habits during the study period, and to take the capsules 30 min before or 120 min after meals, as concomitant ingestion of certain types of food may counteract the effect of flavonoids (Lorenz et al., 2007).

The NC were recruited separately and differed from the participants by being a healthy group that did not receive any intervention, and did not perform cognitive testing. Thus power calculations for further studies might be compromised. However, we provide descriptive statistics including variance measurements which may be helpful in planning sample size in later studies.

An important strength of our study is the well characterized combination of nutraceuticals used, which facilitates comparison with other studies regarding both source of anthocyanins and dosage. Of note, this is, to our knowledge, the first study on cognitive function in adults with increased risk of dementia where Medox^® is being used as the source of anthocyanin. Thus our study might also facilitate investigation of the effect of different proprietary blueberry formulatons, shown by Whyte et al. to be of importance (Whyte et al., 2018).

All things considered, adequately powered, randomized studies are warranted to better understand how anthocyanins and their metabolites may affect relevant mechanisms, including their possible protective role in epigenetic modifications that potentially benefit the aging brain and reduce the risk for dementia.

All participants provided written informed consent, and the study has been approved by the Regional Ethics Committee (Approval 2014/1966). The study has been registered at ClinicalTrials.gov (NCT02409446).

DA, HS, AB, and AL planned and designed the study. AB conducted the study and collected the data. RB performed the serum analyses. MT and AR-M performed the plasma analyses. AB and ID conducted the statistical analyses. All authors wrote the manuscript and critically reviewed the manuscript.

AB has received support for conference participation from Evonik. DA has received research support and/or honoraria from Astra-Zeneca, H. Lundbeck, Novartis Pharmaceuticals, and GE Health, and serves as paid consultant for H. Lundbeck and Axovant. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.