Molecules 27 03737

molecules
Review
Neuroprotective Effect of Caffeine in Alzheimer’s Disease
Y Mukish M Yelanchezian 1 , Henry J. Waldvogel 1 , Richard L. M. Faull 1 and Andrea Kwakowsky 1,2, *
1 Centre for Brain Research, Department of Anatomy and Medical Imaging, Faculty of Medical and Health
Science, University of Auckland, Auckland 1023, New Zealand; [email protected] (Y.M.M.Y.);
[email protected] (H.J.W.); [email protected] (R.L.M.F.)
2 Pharmacology and Therapeutics, School of Medicine, Galway Neuroscience Centre, National University of
Ireland Galway, H91 W5P7 Galway, Ireland
* Correspondence: [email protected]; Tel.: +343-09149-3012
Abstract: Alzheimer’s disease (AD) is the leading cause of dementia, predicted to be the most
significant health burden of the 21st century, with an estimated 131.5 million dementia patients
by the year 2050. This review aims to provide an overview of the effect of caffeine on AD and
cognition by summarizing relevant research conducted on this topic. We searched the Web of
Science core collection and PubMed for studies related to the effect of caffeine on AD and cognition
using title search terms: caffeine; coffee; Alzheimer’s; cognition. There is suggestive evidence
from clinical studies that caffeine is neuroprotective against dementia and possibly AD (20 out of
30 studies support this), but further studies, such as the “ideal” study proposed in this review, are
required to prove this link. Clinical studies also indicate that caffeine is a cognitive normalizer
and not a cognitive enhancer. Furthermore, clinical studies suggest the neuroprotective effect of
caffeine might be confounded by gender. There is robust evidence based on in vivo and in vitro
studies that caffeine has neuroprotective properties in AD animal models (21 out of 22 studies
support this), but further studies are needed to identify the mechanistic pathways mediating
Citation: M Yelanchezian, Y.M.;
these effects.
Waldvogel, H.J.; Faull, R.L.M.;
Kwakowsky, A. Neuroprotective Keywords: caffeine; coffee; cognition; Alzheimer’s disease; dementia
Effect of Caffeine in Alzheimer’s
Disease. Molecules 2022, 27, 3737.
https://doi.org/10.3390/
molecules27123737 1. Introduction
Academic Editor: José Marco- Alzheimer’s disease (AD) is reported to be the leading cause of dementia and the
Contelles significant healthcare burden of the 21st century [1]. In 2015, over 46 million people were
reported to be living with dementia (costing US 818 billion), a figure projected to reach
Received: 20 April 2022 131.5 million dementia patients by the year 2050 [2]. Developed countries with an aging
Accepted: 6 June 2022
population are expected to be worse hit by this burden. Typical clinical presentation of
Published: 10 June 2022
dementia includes memory impairment and executive function decline that interferes
Publisher’s Note: MDPI stays neutral with daily activities making the elderly less independent and forcing them to engage with
with regard to jurisdictional claims in support services [1]. Atypical presentation of dementia consists of a more pronounced
published maps and institutional affil- memory deficit causing language, visual, and executive problems [1]. Atypical dementia
iations. is more common in early-onset dementia, which is reported to have a strong genetic
component [1].
The pathophysiology of AD is based on the accumulation of abnormally folded
Aβ and Tau proteins in amyloid plaques and neuronal tangles that contribute to neu-
Copyright: © 2022 by the authors.
rodegeneration in patients’ brains [3]. Much of this evidence comes from studying
Licensee MDPI, Basel, Switzerland.
familial AD, where there exist mutations in APP genes, which alter the action of γ-
This article is an open access article
secretases that cleaves Amyloid Precursor Protein (APP), causing an accumulation
distributed under the terms and
and aggregation of Aβ peptide [4]. Hyperphosphorylated Tau (PTau) protein, another
conditions of the Creative Commons
Attribution (CC BY) license (https://
prerequisite for AD diagnosis, accumulates intracellularly and fibrillates into paired
creativecommons.org/licenses/by/
helical filaments that form neurofibrillary tangles [5]. It has been proposed that PTau
4.0/).
can further accelerate Aβ dysfunction [5]. A significant genetic factor for AD is APOE
Molecules 2022, 27, 3737. https://doi.org/10.3390/molecules27123737 https://www.mdpi.com/journal/molecules

Molecules 2022, 27, 3737 2 of 41
mutations, with the lifetime risk of AD being 50% for homozygous APOE4 carriers and
20–30% for APOE3 and APOE4 heterozygous carriers [6]. However, even without an
APOE mutation at 85 yr of age, there is an 11–15% risk of developing AD, indicating
there might be a significant environmental component to AD [6]. Recent evidence has
suggested many lifestyle factors: diabetes, diet, socioeconomic status (SES), education,
physical and mental activity, depression, tobacco use, and alcohol intake may affect the
chance of developing AD and dementia [7]. The Rotterdam study even demonstrated
that eliminating the seven most hazardous risk factors could reduce the incidence
of dementia by 30% [8]. In the absence of a definitive clinical treatment currently,
eliminating these modifiable risk factors is our best tool for reducing the burden of
dementia and AD.
Coffee, the most heavily consumed caffeinated beverage, has been a popular research
topic in AD, with several epidemiological studies positing its neuroprotective effect [9]. Cof-
fee comprises a few independently neuroprotective components: caffeine, chlorogenic acid,
caffeic acid, and trigonelline [10]. Chlorogenic acid has been shown to reduce blood-brain
barrier (BBB) damage and improve neuronal differentiation in mice [11]. Caffeic acid and
phytochemicals in coffee act as antioxidative and anti-inflammatory substances, thereby
helping reduce cognitive decline [12]. Trigonelline from coffee beans has been shown to
alleviate neuronal loss by reducing oxidative stress, astrocyte activity, and neuroinflam-
mation while preserving mitochondrial integrity [13]. The neuroprotective properties of
coffee have been heavily linked to its high caffeine content, but this has been difficult to
demonstrate independently through epidemiological studies due to the confounding effect
of other components in caffeinated beverages [9].
Caffeine is a natural trimethyl xanthine alkaloid in which the three methyl groups
are located at positions 1, 3, and 7 (1,3,7-Trimethylxanthine) [14]. Caffeine, a key
psychoactive ingredient in coffee, is a short-acting neurostimulator with known neu-
romodulator effects on the brain by inhibiting phosphodiesterase, mobilizing intracel-
lular calcium, antagonism of adenosine receptors, and modulation of GABA receptor
function [15]. Rodent studies have also reported caffeine can inhibit amylogenic-Aβ
protein production and improve cognition in rodent AD models [16]. Caffeine has
high oral bioavailability, with 99% of caffeine being absorbed from the gastrointestinal
(GI) tract into the bloodstream 45 min after ingestion. A peak plasma concentration of
1–10 µM (0.25–2 mg/L) reached 15–120 min post oral ingestion from a cup of coffee in
humans [17].
However, the results from previous studies are controversial, with some reporting
caffeine to be neuroprotective while others report no effects or even detrimental effects on
cognition. Therefore, this study aims to clarify the impact of caffeine on cognition and AD
by reviewing all the relevant research published on this topic.
2. Methods
In the present review, we searched the Web of Science core collection and PubMed
for studies related to the effect of caffeine and cognition. We used the title search terms
caffeine, coffee, Alzheimer’s, cognition and excluded all reviews.
We did not use the search term tea because it returned many results related to the
neuroprotective action of herbal tea, which was not associated with the action of caffeine,
the main interest of this review. Additionally, Eskelinen et al. reported that tea drinking
in midlife was not correlated with AD development [18]. They posited that this could
be because of tea’s comparatively lower caffeine content than coffee [18]. However, this
could be biased as the study recruited fewer tea drinkers than coffee drinkers [18]. The
average caffeine content per drink was 60.4 +/− 21.8 mg for instant coffee (14-fold range),
80.1 +/− 19.2 mg for brewed coffee (2.8-fold range), and 28.8 +/− 13.7 mg for tea (5.5-fold
coverage) [19].
Molecules 2022, 27, 3737 3 of 41
Through our review we identified 20 clinical studies that supported the notion that
caffeine/coffee exerts a neuroprotective effect on cognition against dementia and AD
([18,20–38]) (Table 1). Our review of the literature also found 10 clinical studies that did not
support the notion of neuroprotective effect of caffeine/coffee [39–48] (Table 1). We also
reviewed 21 animal model studies that showed caffeine as neuroprotective [49–69] (Table 2)
and used them to explain caffeine’s expected mechanism of action. We also identified one
in vitro and in vivo study that did not link caffeine and cognition [70] (Table 2).
3. Clinical Studies Investigating the Relationship between Caffeine and Cognition

3.1. Longitudinal Epidemiological Studies
Cao et al., gathered two cohorts of 124 participants (65–88 yrs old) and measured
their plasma caffeine concentration as well as their initial neurological status (based on
clinical history, clinical dementia rating (CDR), Multi-Mental State evaluation (MMSE),
psychiatric evaluation, Three-trial Fluid Object Memory Evaluation (TFOME), Hop-
kins’s verbal Learning Test Revised (HVLTR), MRI-volumetric protocol and National
Alzheimer’s Disease and Clinical Centre (NACC) protocol [20]. Participants were
grouped into three categories of cognitive functions: Normal (M), Mild cognitive impair-
ment (MCI), and Dementia (DEM); then, the participants were followed up for a period
of 2–4 yrs, and their cognition was reassessed (same protocol) [20]. This study found that
participants who demonstrated a cognitive decline from initial MCI to DEM had a signif-
icantly lower plasma caffeine concentration (by 51%) than participants who maintained
the level of cognitive impairment (stable MCI) [20]. Furthermore, none of the subjects
with a critical plasma caffeine concentration (>1200 ng/mL) converted to DEM, and
half of stable MCI had plasma concentrations greater than this critical value [20]. Fred-
holm et al. 1999, found that this plasma concentration of coffee (1200 ng/mL or 6 µM)
is the typical plasma caffeine concentration several hours after ingestion of 1–2 cups of
coffee after decreasing from a peak of 10 µM–20 µM an hour after coffee ingestion as
reported by Culm-Medrek et al., 2005 [71,72]. Cao et al., also measured 11 cytokines
and found that 3 (GCSF, IL-10, and IL-6) were lower in participants who had a cognitive
decline from initial MCI to DEM [20]. The main strength of this study is that it uses
plasma caffeine concentration which is a more objective measure than using a recall
dietary survey of caffeine intake as used in other studies [20]. However, this approach
is also associated with a few limitations, like only one measurement of plasma caffeine
levels was obtained; thus, it does not allow detection of a change in caffeine habit [20].
Also, the study did not account for other confounding variables, i.e., lifestyle choices.
The follow-up period of 2–4 yrs may be too short to determine if caffeine-cognition has a
true causal relationship [20].
Another study recruited 1445 cognitively normal subjects screened with: the Bab-
cock Story Recall Test (BRST), Activities of Daily Living (ADL), and MMSE, then
grouped them into two groups: rarely consumed coffee (<1 cup/day) and habitual
moderate coffee drinker (1–2 cups/day) based on Food Frequency Questionnaire (FFQ)
results; subjects were followed up for a 3.5 yrs median and incidence of developing MCI
based on MMSE, GDS was recorded [21]. Solfrizzi and colleagues found that habitual
moderate coffee drinkers (HR:0.31, 95% CI:0.13–0.75) had a lower incidence of MCI than
rare coffee drinkers (HR:0.47, 95% CI:0.211–1.02); this minor difference by hazard ratios
of 0.16 indicates that there is only a slight decrease in risk of dementia among coffee
consumers [21]. The study also found that those who altered their coffee habit had
an increased risk of MCI than those with a constant coffee habit: increased by >1 cup
(HR:1.8, 95% CI:1.11–2.92), decreased by <1 cup (HR:2.17, 95% CI:1.16–4.08) this shows
that there is almost doubling of dementia risk when there is a change in the coffee
habit [21]. Furthermore, high coffee habit >2 cups/day had no correlation with MCI
incidence compared to those who rarely consumed coffee (HR:0.26, 95% CI:0.03–2.11),
as shown by the large CI [21]. A significant strength of this study is that it showed the
effect of changing coffee habits on the risk of developing MCI [21]. However, this study
Molecules 2022, 27, 3737 4 of 41
has a few limitations like recall bias as a coffee habit was based on FFQ, underestimation
of caffeine intake as it does not include other sources of caffeine (tea, chocolate,) and a
short follow-up period [21].
As part of a larger study, Driscoll et al., recruited 6467 participants from Women
Health Initiative (WHI) Hormonal therapy RCT trial and determined their self-reported
caffeine intake at enrolment using FFQ [22]. Then the subject’s cognitive function was
screened annually for ten years or less over the telephone using a 100-point Modified
Mini-Mental State (3MS) exam and a 40-point Telephone Interview for Cognitive Status-
modified (TICSM); those showing cognitive decline were followed up by board-certified
psychiatrists using the Dementia Questionnaire (DQ) to verify MCI [22]. Then propor-
tional hazard regression (HR) was performed to assess the risk of developing MCI based
on their baseline caffeine intake while adjusting for risk factors: such as hormone ther-
apy, age, race, education, BMI, sleep quality index, depression, hypertension, diabetes,
history of cardiovascular disease (CVD), smoking and alcohol consumption [22]. Driscoll
et al., found that women consuming above the median (175 mg/d) levels of caffeine
(mean intake = 261 mg for this group) had only a small effect on lowering the risk of
developing dementia by 26% or any cognitive impairment (HR = 0.74, 95% CI:0.60–0.91)
compared to those consuming below median levels (mean intake = 64 mg for this
group) [22]. This study also reported that a cup of 8-ounce black coffee contained 95 mg
of caffeine; thus, the median caffeine intake is equivalent to 1.8 cups of black coffee [22].
A significant strength of this study is that the subjects were extensively screened and
characterized through a detailed prospective follow-up study, which helped adjust for
confounding variables and a short lag between screens which helped provide a more
sensitive record of MCI onset [22]. However, this study has limited generalizability as
only older post-menopausal women who tended to be better educated were included in
the study [22].
Paganini-Hill et al., recruited 587 subjects >90 yrs old (mean = 93 yrs) who showed
no signs of dementia at enrolment, and their lifestyle factors data (smoking, alcohol,
caffeine intake, vitamin supplement, and exercise) was also collected at enrolment, and
20 yrs previously from the Leisure World Cohort health survey (1981–1985) [23]. Then the
participants were followed up for 36 months, and their cognitive status was determined
by trained health professionals using a battery of neuropsychological tests: neurological
exam, MMSE, informant questionnaire, DQ, and CASI-short [23]. This study found that
those who consumed >200 mg/day of caffeine had a significantly lower risk of about
34% (HR = 0.66, p < 0.05) of dementia than those who consumed <50 mg/day of caffeine
by using Cox regression analysis [23]. A strength of this study is that it prospectively
studied cognitively normal individuals and routinely examined their cognitive status with
a short lag time in-between, which allowed early identification of cognitive decline and
dementia [23]. However, this study has some severe limitations as it only assessed lifestyle
factors from 2-time points 20 yrs apart. It did not account for habit changes, and the study
subjects were also predominantly moderately affluent Caucasians, limiting the study’s
generalizability [23].
In a large-scale study, 13,137 cognitively healthy subjects were recruited (>65 yrs)
from the Ohsaki Cohort study 2006, FFQ assessed their coffee intake, and other lifestyle
factors were also simultaneously collected and adjusted for: baseline age, BMI, green
tea consumption, education, CVD history, fractures, stroke, diabetes, smoking, alcohol
intake and social support [24]. The subjects were followed up for 5.7 yrs, and incidents of
dementia were calculated based on data from the Long-Term Care Insurance database [24].
A multivariate analysis found that coffee consumption and incidence of dementia for the
categories: never, occasionally, 1–2 cups/day, ≥3 cups/day had a hazard ratio of 1.00, 0.73
(95% CI: 0.62–0.82), 0.72 (95% CI: 0.61–0.84) and 0.82 (95% CI: 0.65–1.02), respectively. These
findings indicated that 1–2 cup and ≥3 cups was moderately neuroprotective, reducing by
28% and 18%, respectively [24]. This protective effect of caffeine was more significant among
women, non-smokers, and non-drinkers groups than in the studies’ general population.
Molecules 2022, 27, 3737 5 of 41
Compared to the previous study, a substantial benefit of this study was its very large sample
size and multivariate adjustment for confounders, which included social support [24].
However, this study did have a few limitations in assessing coffee intake: it did not
measure coffee intake in midlife, did not assess if there was a change of habit after baseline,
and did not differentiate between decaffeinated caffeinated coffee [24]. This study also
may have suffered from some reverse causality due to a lack of data sensitivity as they
did not exclude those who showed a cognitive decline at baseline but were not certified
disabled [24].
Haller et al. recruited 45 elderly controls and 18 MCI who were chronic coffee con-
sumers (1–3 cups/day) [25]. They were followed up for 18 months, and their cognitive func-
tion was assessed (MMSE), which was used to group them into Stable-controls cognitive
(24-sCON), mild cognitive impairment (18-MCI), and deteriorating-controls (21-dCON) [25].
The participants were put on caffeine detox for 18 h and then given 200 mg caffeine or a
placebo 30 min before being subjected to an n-back task (established WT fMRI test) [25].
The MR data were analyzed with: a hypothesis-driven general linear model (GLM) analy-
sis of task-related activation, tensorial induced component analysis (TICA) of functional
connectivity, arterial spin labeling (ASL) perfusion, Gray matter voxel-based morphometry
(VBM), and white matter DTI tract-based special statistics (TBSS) [25]. Haller et al., found
no difference in working memory (assessed by fMRI n-back tasks) performance between
sCON and dCON, while MCI was less accurate and slower (p < 0.05) [25]. Furthermore, the
dCON group also had a less pronounced acute caffeine-induced brain activation, which
was restricted to the right hemisphere (p < 0.05) and reduced caffeine-induced Default
Mode Network deactivation compared to sCON (p < 0.01) [25]. This decreased sensitivity
of caffeine effects in dCON is in line with the idea caffeine is a cognitive normalizer instead
of a cognitive enhancer, and complex fMRI patterns are possibly due to existing functional
changes despite behavioral pattern maintenance [25]. Strengths of this study were the
exclusion of potential confounding alteration in GM by VBM and WM by DTI TBSS; and
using ASL to measure brain perfusion, which confirmed that even though there was a
generalized 25–30% decrease in CBF, there was not any localized change in CBF, ensuring
that fMRI changes are not due to global CBF differences. This study also has several
limitations, such as a small sample size and the inability to comment on long-term changes
in brain activation patterns induced by caffeine [25]. Then there is also the possibility of
non-excludable de-novo brain pathologies during follow-up [25].
In another study, Haller et al., recruited 145 cognitively stable elders screened by
cognitive test (MMSE, HAD, LIDA) and used a self-administered questionnaire to assess
their coffee, chocolate, and wine consumption [26]. They were followed up for 3 yrs,
during which MRI imaging and two neuropsychological examinations were administered
testing: attention, working, digit episodic, executive, language, visual, phenomics verbal
fluency, praxis ideomotor, reflexive, and constructive memory [26]. These data were used to
group participants into stable-cognitive (52-sCON), intermediate-cognitive (62-iCON) and
deteriorating-cognitive (32-dCON). The MRI data were analyzed by: whole-brain VBM,
ASL, diffusion tensor imaging TBSS, and GM region of interest (ROI) analysis [26]. This
study found that moderate coffee consumers are less likely to be categorized as dCON
(ORadjusted : 0.447, 95% CI:0.210–0.952, p = 0.037) [26]. Moreover, MRI imaging found a
negative correlation between VBM and caffeine only for sCON, notably in the WM (left
parietal and right frontal), indicating fewer WM lesions and increased cerebral blood flow,
but there was no association among iCON and dCON [26]. This positive relationship
between cognition and caffeine only being present in sCON further supports the notion
that caffeine is a cognitive normalizer and not a cognitive enhancer [26]. The strength of
this study includes longitudinal follow-up and a lack of health-related confounders [26].
However, there were also a few limitations, like the study subject being adults who lack
vascular pathology, limiting the study’s generalizability among the general population
with CVD pathology [26].
Molecules 2022, 27, 3737 6 of 41
West et al., recruited 638 elderly (+65 yrs) subjects with Type 2 diabetes (T2D) from
the Israel Diabetes and Cognitive Decline study (IDCD). The subject’s caffeine intake
was recorded using FFQ [33]. Their cognition was measured using a neuropsychological
test battery that looked at four factors: episodic memory, executive function, semantic
categorization, and working memory [33]. A further subgroup of subjects (185) was
randomly selected and subjected to MRI imaging to estimate WM and GM volumes [33].
West et al., using linear regression adjusting for cognitive-related covariates (SES, diet,
cardiovascular, thyroid, and type of T2D), found that higher caffeine intake was related
to better overall cognition (p = 0.018), working (p = 0.002), executive function (p = 0.047),
semantic memory (p = 0.026). This effect on cognition was amplified in the older group
(above median) compared to the younger [33]. The MRI imaging also found that higher
caffeine intake resulted in higher GM volume (B = 0.198, p = 0.033) [33]. Which indicates
reduced neuronal death and a possible mechanism of neuroprotective action of caffeine [33].
The strength of this study is that it uses a large study sample size and a unique study
population (T2D) while adjusting for potential confounding [33]. However, this also acts
as a limitation as it makes it difficult to generalize this study’s findings to the non-T2D
population [33].
Vercambre et al., recruited 2475 elderly (+65 yrs) women from the Women’s Antioxi-
dant Cardiovascular Study (WACS), and their caffeine intake at baseline was assessed by
116 item-food frequency questionnaires [27]. Then their global cognition was assessed using
four Telephone Interview of Cognitive Status (TICS) at 2-year intervals. Vercambre et al.,
reported consumption of caffeinated coffee was correlated with significantly slower rates
of cognitive decline (p = 0.05) but not for other caffeinated products, i.e., tea, chocolate, and
cola [27]. The rate of difference between the highest and lowest quantile of caffeine intake
(>371 and <30 mg/day) and cognitive decline was equivalent to that of 7 yrs apart in age
(p = 0.006) [27]. The strengths of this study were that it was a longitudinal study, adjusting
for confounders, and included a large sample size [27]. However, there were also several
limitations: using a self-reported questionnaire to assess caffeine intake increases the risk
of recall bias, the study only looked at caffeine intake at baseline, which might not reflect
long term use, and this study focused on caffeine as the sole active component in coffee,
but other studies have reported that polyphenols in coffee might exert an independent
protective effect against cognitive decline [27].
Eskelinen et al., randomly recruited 1409 individuals (followed up for 21 (median = 4.9)
years) from the FINMONICA study (1972, 1977, 1982, 1987), which examined lifestyle, diet,
BMI, BP, health status, and serum cholesterol through a self-administered questionnaire [18].
During the follow-up in 1998, the previous survey method was readministered, and ApoE
genotyping and cognitive screening using MMSE, DSM-IV, and NINCDS-ADRDA [18].
The study reported that coffee drinkers in midlife had a markedly lower risk of developing
dementia and AD even after adjustment [18]. The lowest risk of dementia (OR: 0.34,
95% CI = 0.16–0.73) and AD (OR: 0.38, 95% CI = 0.17–0.89) was found among chronic
coffee consumers (3–5 cups/day) at midlife indication there is about a 66–62% reduction
in risk [18]. Strengths of this study include a long follow-up period, validated protective
effects of caffeine against cognitive decline even after adjusting for hyperlipidemia, and
looking at coffee consumption in midlife using a validated questionnaire [18]. Limitations
of this study include the possibility of recall bias due to the self-reported questionnaire
used and a sample size that is too small to detect possible dose-response effects [18].
Ritchie et al., recruited 4197 women and 2820 men, and their coffee consumption (ques-
tions in the standardized interview) as well confounding factors (age, gender, BMI, educa-
tion, diet, lifestyle, medical history, alcohol, and tobacco intake) were assessed at baseline,
and a 2- and 4-year follow-up [28]. The subject’s cognition was assessed with the MMSE,
Benton Visual retention test (BVRT), Issacs set test (IST), and DSM-IV at baseline and subse-
quent follow-ups [28]. Ritchie et al., multivariate mixed models and multivariate-adjusted
logistic regression indicated that women with higher caffeine consumption (>3 cups/day)
showed less decline in verbal retrieval (OR = 0.67, CI = 0.53–0.85) and visuospatial memory
Molecules 2022, 27, 3737 7 of 41
(OR = 0.82, CI = 0.65–1.03) over 4 yrs than women consuming one cup/day or less [28].
This protective effect was enhanced with age by 43%, 65–74 yrs (OR = 0.73, CI = 0.53–1.02)
and 80+ yrs (OR = 0.3, CI = 0.14–0.63), indicating that the neuroprotective effect of caffeine
is significantly confounded by age in women [28]. There was no relation between cognitive
decline and caffeine intake in men [28]. The strengths of this study are a longitudinal design,
a large sample size, and validated methods of measuring cognition [28]. Limitations of this
study are selective attrition could have promoted a healthy survivor effect among subjects,
and a follow-up period longer than 4 yrs may be necessary to evaluate risk and benefit
adequately [28].
Gelber et al., recruited 3734 Japanese Americans, their coffee intake was assessed
through 24-h dietary recall, and their cognition was screened using 100-point Cognitive
Abilities Screening Instrument (CASI) [39]. A subgroup of 418 participants was subjected
to autopsies where their brains were studied for neuropathological lesions (Alzheimer’s
lesions), microvascular ischemic lesions, neocortical Lewis bodies, hippocampal sclerosis,
and generalized brain atrophy) [39]. The study reported no association between coffee
intake in midlife and risk of cognitive impairment [39]. However, it did find that the highest
quartile of coffee intake (≥411.10 mg/day) was less likely than the lowest quartile of coffee
drinkers (≤137.0 mg/day) to have any type of brain lesion in postmortem examination (OR:
0.45, 95% CI = 0.23–0.89, p = trend 0.04) [39]. The strengths of this study are it measures
coffee intake in midlife, has a large sample size, and is the first study to investigate the
link between coffee intake with postmortem lesions [39]. The limitation of this study
is that it only includes men from a particular ethnic and social group which limits the
generalizability of the findings [39].
Araujo et al., recruited 2914 participants (59 ± 7.2 yrs.) and at baseline, performed
coffee consumption analysis (FFQ), brain MRI and cognitive test battery (Letter digit sub-
stitution task (LDST), Stroop test, Word fluency test (WFT), 15-word learning test (WLT)
and Purdue pegboard (PBB) [40]. Then the subjects were followed up for 5 yrs, and the
baseline analysis was repeated [40]. The cross-sectional segment of this study at base-
line reported higher caffeine intake was associated with minor reduction in prevalence of
lacunar infarcts (OR per cup = 0.88, 95% CI = 0.79–0.98), smaller hippocampus volume
(diff size = −0.01, 95% CI = −0.02–0.00), small improvement in cognitive performance in:
LDST (difference = 1.13, 95% CI = 0.39–1.88), WFT (difference = 0.74, 95% CI = 0.04–1.45),
Stroop test (1.182, 95% CI = 0.23–3.41), and worse cognitive performance in WLT (−0.38,
95% CI = 0.74–0.02) but this minor reduction of risk might not result in noticeable changes
by the person [40]. Furthermore, after the five-year repeat cognitive assessment, the associ-
ation between higher caffeine intake and improved cognition was not found. The study
hypothesized that this discrepancy could be because caffeine is a short-term neurostim-
ulator and long-term exposure to caffeine causes the body to build up a tolerance, and
thus the effect of caffeine wanes [40]. However, these relationships were not found when
followed up longitudinally [40]. Strengths of the study include having a population-based
design with large sample size and using brain MRI to measure cognitive function in the
same population. The limitation of the study is that it did not look for a change in the coffee
habit during the follow-up period [40].
Mirza et al., recruited 4368 subjects from the Rotterdam study; their coffee consump-
tion was measured as part of a home interview, covariate data (BMI, age, health, smoking,
alcohol, lifestyle) was gathered from the Rotterdam Study, and cognition was assessed
through MSSE, geriatric mental schedule (GMS), and Cambridge examination for mental
disorders among the elderly (CAM-DEX) [41]. The study data was stratified into short
follow-up (0–4 yrs) and long follow-up (>4 yrs) till 21 yrs [41]. Mirza et al., reported
that during short follow-up, they found those who consumed >3 cups/day had a 30%
lower risk of developing dementia (HR = 0.70, 95% CI = 0.51–0.96) than those who drank
<1 cup/day [41]. However, this relationship did not extend to the long follow-up [41]. The
study reported this could be subject to confounding as the subject reported that chronic
coffee consumers are also more likely to have a poor lifestyle with a high lipid diet which is
Molecules 2022, 27, 3737 8 of 41
a risk factor for dementia [41]. Strengths of this study include having a large sample size,
long follow-up with stratification, and an intensive dementia case verification protocol [41].
Limitation of this study: coffee intake was based on dietary recall increasing the risk of
recall bias, and other confounding factors were not accounted for [41].
Arab et al., recruited 4809 participants (>65 yrs) from the cardiovascular health study
and followed them up for a median of 7.9 yrs [29]. The subject’s intake of caffeinated
beverages was assessed using FFQ, and their cognition was measured annually using
Modified mini-mental state examinations [29]. The study adjusted for confounding (age,
education, SES, depression, APOE genotype, medical history, smoking) and used linear
mixed models to analyze the data [29]. This study found that in fully adjusted models, the
intake of coffee and tea modestly reduced rates of cognitive decline in some but not all
women, and there was no dose-effect relationship among the women [29]. No consistent
effect was identified for men [29]. The strengths of this study are that it has a large follow-
up period and uses a validated measure of cognition [29]. The limitation of this study is
that FFQ did not specify the amount of beverage but only its frequency which reduces the
accuracy of the dose-response findings [29].
Fisher et al., recruited 2622 (75+ yrs.) participants from the German study on aging,
cognition, and dementia (AgeCoDE) whose food intake was measured using a single-food
questionnaire (wine, coffee, green tea, olive oil, fresh fruits, vegetable, and red meats) and
their incidence of dementia and AD assessed through (CERAD and SIDAM) over 10 yrs.
was recorded [42]. The data was then analyzed using multivariate-adjusted joint modeling
considering gender and Apolipoprotein E4 (APOE e4) [42]. Fisher et al., did not find
any statistically significant association between coffee consumption and the incidence of
dementia/AD [42]. The strength of this study was the inclusion of dropout time to allow an
unbiased account of cognitive decline in the presence of missing data [42]. The limitation
of this study was that dietary recall was only done at baseline, and they did not measure
for change in the dietary habit over follow-up [42].
Santos et al., included 648 participants (≥65 yrs), and assessed their baseline (1999–2003)
caffeine intake using FFQ and cognition using MMSE [30]. They were then followed up,
and their cognition was assessed again using MMSE (2005–2008) [30]. The data was then
adjusted for confounding factors: age, education, gender, smoking, alcohol, BMI, hyper-
tension, and diabetes [30]. Santos et al., reported that those with the third quartile of
caffeine intake (>62 mg/day) had a significantly lower risk of cognitive decline by 51%
(RR = 0.49, 95% CI = 0.24–0.97) compared to those with the first quartile of caffeine intake
(<22 mg/day) among women only [30]. This reduction is expected to result in noticeable
benefits for women. No statistically significant relationship between caffeine intake and
cognition was found among men [30]. The limitation of this study was that it did not
account for incomplete follow-up as only 58.2% of the initial cohort completed the study;
therefore, there could be selection bias that limits the study’s generalizability [30].
3.2. Cross-Sectional Studies

Cornelis et al., recruited 445,786 participants (37–73 yrs) from 22 biobank centers
across the UK [45]. The subjects filled out an extensive questionnaire that contained
caffeinated beverage intake (used to calculate average daily caffeine), medical history,
lifestyle, and diet [45]. Participants’ cognition was screened for prospective memory (PM),
pairs matching (Pairs), Symbol Digit Substitution (SDS), fluid intelligence (FI), and reaction
time (RT) using a computerized cognitive function test [45]. The data was then analyzed
using multivariate analysis to identify interactions between coffee, tea, and genetic-based
caffeine metabolism score (CMS) on cognitive function [45]. Cornelis et al., reported that
coffee intake (≥1 cup) significantly decreased reaction time, pairs matching, Trail making
test B, and symbol digit substitution. No, statistically significant relationship was identified
between cognitive function and CMS × tea, CMS × coffee, and CMS × caffeine [45]. The
strength of this study was its large sample size and its ability to adjust for genetic caffeine
metabolism. Limitations include the possibility of unmeasured confounding, and the study
Molecules 2022, 27, 3737 9 of 41
sample also suffers from “healthy volunteer’ selection bias and may not represent the wider
population [45].
Another study by Cornelis et al., recruited 434,900 participants (37–73 yrs) from
22 biobank centers across the UK [46]. The subjects filled out an extensive questionnaire
that contained medical history, lifestyle, and diet [46]. Recent caffeine intake (last hour) was
recorded during the physical assessment, where participants completed at least one out of
four self-administered cognitive function tests: prospective memory (PM), pairs matching
(Pairs), fluid intelligence (FI), and reaction time (RT) [46]. The data was then analyzed using
multivariate analysis to identify interactions between recent caffeine intake, genetic-based
caffeine metabolism score (CMS), and cognitive function [46]. Cornelis et al. [46] reported
that among white participants recent coffee consumption was correlated with higher RT
performance but worse FI, Pairs, and PM (p ≤ 0.004) [46]. Among non-white participants
similar associations were found FI (p = 0.09), Pairs (p = 0.03), and PM (p = 0.34) [46]. The
limitations of this study are that yes/no self-reported recent caffeine consumption, which
suffers from recall bias, was used instead of a biomarker, and information regarding caffeine
source, amount, or preparation was not considered [46].
Ritchie et al., recruited 1193 elderly (+65 yrs), including those with depressive symp-
tomology and T2D [43]. The subject’s caffeine intake was recorded at baseline during
the interview, along with their cognition (MMSE), serum glucose, and B-amyloid levels
(known memory confounders) were also recorded [43]. Higher caffeine intake was linked
to a significant decrease in incidental diabetes in men (HR:0.64, 95% CI:0.42–0.97) and a
significant increase in incidental diabetes risk in women (HR:1.51, 95% CI:1.08–2.1), no
statistically significant association was found between caffeine and depression or Aβ lev-
els [43]. The study also did not find that caffeine was neuroprotective against dementia
among women [43]. The study found no evidence that decreased risk of dementia among
heavy caffeine-consuming women was confounded by diabetes or depression [43]. A
limitation of this study, like many others, is that it assumes that caffeine is the only neu-
roprotective substance in tea or coffee, but Alves et al., have reported on the estrogenic
properties of coffee through its high isoflavone content, which could independently exert
neuroprotective properties [43,73].
Kim et al., recruited 411 subjects and screened them using the CERAD-K neuropsycho-
logical examination into cognitive normal (CN = 282) and MCI = 129 [31]. The participant’s
coffee consumption (current and lifetime) was grouped into the low coffee intake (<2) and
high coffee intake (≥2). Then the subjects underwent PET and MRI scans to measure cere-
bral Aβ deposition, AD-CM, AD-CT, and WMH [31]. This study found that higher coffee
intake was significantly associated with lower Aβ positivity than low coffee consumption,
even after adjusting for confounding factors [31]. However, current nor lifetime coffee
intake was associated with hypometabolism, AD-signature region, and WMH volume [31].
This absence in a change of WM volume contrasted with MRI studies by Ritchie et al., and
Haller et al., who found caffeine decreased the amount of WM lesion/cranial volume in
cognitively stable elders [26,35]. The strength of this study is that it is the first study to
investigate the association between coffee intake and in vivo AD pathologies [31]. However,
it also has a few limitations; since it is a cross-sectional study, it cannot establish a causal
relationship between coffee and cognition. Since coffee intake was based on recall, there is
an increased risk of recall error [31].
A cross-sectional study by Iranpour and colleagues recruited 1440 adults (>60 yrs)
from the National Health and Nutrition Examination Survey (NHANES) [34]. Twenty-
four-hour dietary recall data assessed the caffeine intake, and cognition was measured
using CERAD and DSST. They also collected covariate data on age, sex, SES, lifestyle,
BMI, and history of the disease. The study found that the highest quartile of caffeine
intake was positively associated with better cognitive function in the crude model with a
p < 0.05, which means the relationship is statistically significant [34]. After adjusting for
confounding, the association was only marginally significant in the CERAD word recall
test (p = 0.09), and this trend was enhanced among men (B = 0.001, p = 0.004) but not
Molecules 2022, 27, 3737 10 of 41
females (B = 0.00007, p = 0.89) [34]. This contrasts with the other three previous studies
that found that the neuroprotective effect of caffeine was enhanced among women and
not males [24,28,35]. This clearly, indicates that more studies are needed to comment on
how the neuroprotective effect of caffeine is affected by gender. The strengths of this study
are that it had a large sample size and accounted for various confounders [34]. However,
this study also had a few limitations: being a cross-sectional study, it suffers from reverse
causality and using dietary recall, which increases random measurement error [34].
Ritchie, et al., recruited 641 elderly (>65 yrs) persons and recorded their caffeine intake
as well as confounding factors (SES, education, mobility, BMI, alcohol intake, smoking,
disease) at baseline using a questionnaire [35]. Then the participants underwent MRI imag-
ing to estimate white matter lesions (WML) and white matter volume (WM) volume [35].
Ritchie et al., found that the mean log-transformed WML/cranial volume ratio after ad-
justing for women who consumed more than three units of caffeine (−1.23, SD = 0.06)
was significantly lower than women who consumed two to three units of caffeine (−1.04,
SD = 0.04) or one unit or less (−1.04, SD = 0.07) [35]. The study also showed increased
cerebral perfusion in chronic coffee consumers, indicating a possible neuroprotective mech-
anism of coffee [35]. However, this relationship did not extend to the male population, who
had no statistically significant association between caffeine and WML [35]. A strength of
this study was that it used large epidemiological data and considered multiple confounding
factors [35]. However, the limitations of this study, like many done on this topic, was that it
did not consider lifetime caffeine intake, and self-reported caffeine intake was used, which
is prone to recall bias. Furthermore, MRI imaging was performed only once, so the study
could not describe the neurological changes over time induced by caffeine intake [35].
Kyle et al., recruited 351 participants (64 yrs) born in 1936 and sat Moray House
Test (MHT) [44]. The subjects underwent an interview with a health professional to
extract information (medication, lifestyle, SES, gender, age, disease history) and had their
cognition tested with MMSE [44]. The participant’s caffeine intake was also measured
using a MONICA food frequency questionnaire [44]. The study found that caffeine intake
was correlated with a slower digit symbol (F = 3.38, p < 0.02), but this was removed after
accounting for SES [44]. Kyle et al., reported that once adjusted for confounding and SES,
there was no evidence that caffeine affected cognition [44]. A limitation of this study is it
used a self-reported questionnaire which increases the chance of recall bias, and the study
also has a small sample size [44].
Dong et al., recruited 2513 participants (≥60 yrs) from the National Health and Nu-
trition Survey (NAHNES) [36]. Coffee and caffeine intake was recorded using two 24-h
dietary recall questionnaires, and cognition was assessed using the CERAD test, animal
fluency test, and DSST [36]. Dong et al., using binary logistic reasoning and restricted cubic
spline models, reported that those with 226.4–495 g/day caffeine intake had a significantly
better performance by 44% (OR = 0.56, 95% CI = 0.35–0.89) on DSST compared to those
who reported no caffeine intake [36]. Furthermore, those who reported ≥384.8 g/day also
had moderately better performance (OR = 0.68, 95% CI = 0.48–0.97) compared to the lowest
quartile of caffeine intake and (OR = 0.62, 95% CI = 0.38–0.98) for CERAD [36]. A positive
association was reported between caffeine/coffee intake and CERAD and DSST Score
but no association between decaffeinated coffee and cognition [36]. The strength of this
study was that it used a large nationally (USA) representative sample of older adults in the
study [36]. However, as this is a cross-sectional study, it is more prone to reverse-causality
and cannot confirm causality [36].
Additionally, a few cross-sectional studies did not directly examine the effect of caffeine
on cognition. Al-Khateeb et al., studied the effect of serum copper/lipid on cognition and,
incidentally, found that increased coffee intake demonstrated a 6.25-fold lower risk for
cognitive decline [32]. Furthermore, Kim et al., and Hosking et al., examined the effect of
caffeine on cognition as part of a specific diet and found that the coffee diet was correlated
to worse cognitive performance [47,48]. However, these studies had a severe limitation as
Molecules 2022, 27, 3737 11 of 41
coffee was considered part of a diet that contained many components (i.e., high fat), posing
a higher risk of cognitive decline [47,48].
3.3. Randomized Control Studies

Haler et al., performed a double-blind placebo-controlled fMRI study during an
n-back working memory task in 17 individuals with MCI (70.7 ± 4.6 yrs) and 17 age-
matched healthy individuals (HC) (68.3 ± 2.8 yrs) who were cognitively screened (MMSE,
Hospital Anxiety and Depression Scale (HAD), Lawton’s instrumental daily activities
(LIDA)) [38]. The MCI was then further described with neuropsychological examina-
tions, which tested: (attention, working, digit, episodic, executive, language, visual,
phenomics verbal fluency, praxis ideomotor, reflexive, and constructive) memory [38].
All subjects were chronic coffee consumers (1–3 cups/day) who were detoxed from
caffeine for 18 h and given 200 mg of caffeine or placebo tablets 30 min before neu-
roimaging [38]. The scan data was assessed to measure behavioral data (SPSS statistics),
General Linear Model (GLM) analysis of task-related activation, Tensorial-independent
component analysis (TICA) analysis of functional connectivity, analysis of Atrial Spin
Labelling (ASL) perfusion, grey matter voxel-based morphometry (VBM) analysis, and
white matter microstructure Track based spatial statistics (TBSS) fractional anisotropy
analysis [38]. The study reported that acute caffeine administration induced a more
prefrontal activation in HC and a more diffuse posteromedial activation in MCI [38]. In
MCI, TICA documented significant-caffeine related enhancement in the activation of the
prefrontal cortex, supplementary motor areas, ventral premotor, parietal cortex, basal
ganglia, and cerebellum compared to HC [38]. This suggests the posterior displacement
of working memory-related brain activation patterns after caffeine administrations in
MCI represents a compensatory mechanism to counterbalance frontal lobe dysfunc-
tion [38]. This also adds to the evidence that caffeine acts as a neuro normalizer instead
of a neuroenhancer [38]. As previously suggested by longitudinal epidemiological stud-
ies, coffee might be a cognitive normalizer [25,26]. Additionally, West et al. found that
caffeine’s effect on cognition was enhanced among the older group [33]. The absence of
a significant difference in ASL signifies a neuronal difference rather than a purely perfu-
sion difference. The exclusion of potentially confounding differences in VBM and TBSS
analysis between HC and MCI reduces confounding due to differences in Grey-matter
densities and white matter microstructure [38]. The strength of this study is that it uses
both task-related model-driven and independent component data-driven analysis of
fMRI while controlling for the direct vasoconstrictive effect of caffeine [38]. An obvious
limitation of this study is the inability to comment on the long-term effects of caffeine,
like stabilization of the Blood-Brain Barrier (BBB) [38].
Lin et al., recruited 20 healthy (good sleep patterns, no substance use, young (18–35),
BMI (18–25)) male participants who were habitual coffee drinkers [37]. They were placed
in a 9-day ambulatory phase during which ten were placed on (3 × 150 mg/day caffeine),
and the other ten were given a placebo (3 × 150 mg/day mannitol) [37]. Then on the
10th day, subjects were woken at normal time and subjected to N-back task MRI imaging
and EEG 12.75 h. after awakening and 5.5 h. after the last caffeine treatment [37]. The
imaging data were analyzed for Grey Matter Volume (GMV) and cerebral blood flow
(CBF) [37]. In this study, higher caffeine intake was associated with reduced GMV in the
medial temporal lobe compared to placebo, even after adjusting for increased CBF induced
by caffeine [37]. Caffeine treatment was also associated with poor working memory but
not sleep quality [37]. The strength of this study is that, being an RCT and it managed
to significantly reduce environmental confounding factors [37]. This study has some
limitations, such as the small sample size and that the data might be affected by genetic
caffeine insensitivity [37].
Molecules 2022, 27, 3737 12 of 41
Table 1. Overview of clinical studies that investigates the relationship of caffeine and cognition in dementia.
Participants Treatment
Study Study Design Cognition Main Outcomes
Population Groups Size Age (yrs.) Caffeine Follow Up Tests
• subjects with cognitive
decline during follow up
• Initial normal remained (MCI > DEM), had
Normal during follow • clinical significantly lower baseline
up (n = 60) history plasma caffeine concentration
• Initial normal but • psychiatric than participants who
declined to MCI during 124 subjects’ total evaluation maintained their level of
Longitudinal follow up (n = 9) Tampa cohort Baseline Plasma Between 2–4 yrs. • MRI cognitive impairment
[20]
epidemiological • Initial MCI and (n = 81) 65–88 yrs caffeine concentra- Average • CDR (stable MCI)
(Cao et al., 2012)
study maintained MCI Miami cohort tion measured 2 12 –3 yrs • MMSE • a critical baseline plasma
(n = 21) (n = 43) • TFOME concentration of 1200 ng/mL
• Initial MCI declined to • HVLTR was identified
DEM (n = 11) • NACC • out of 11 cytokines measured,
• Initial DEM maintained protocol tests 3 (GCSF, IL-10, and IL-6) were
DEM (n = 23) lower in participants who
experienced cognitive decline
from initial MCI to DEM
• habitual moderate coffee

drinkers had a lower risk of
• rarely consumed coffee
developing MCI than those
(0–1 cup/day),
who rarely drank coffee
(n = 886)
• those who altered their coffee
• moderate levels of
[21] Longitudinal 1445 cognitively FFQ at consumption habit had an
coffee consumers • BRST
(Solfrizzi et al., epidemiological normal at 65–84 yrs 1st = 1992–1993 Median 3.5 yrs increased risk of MCI than
(1–2 cups/day), • ADL
2015) study baseline subjects 2nd = 1995–1996 those with a constant
(n = 409) • MMSE
coffee habit
• higher level of
• there was no MCI incidence
coffee consumers
correlation between those
(>2 cups/day), (n = 150)
with higher levels of coffee
consumption and those who
rarely consumed coffee
Molecules 2022, 27, 3737 13 of 41
Table 1. Cont.
• <75 mg caffeine • median caffeine intake was
(n = 1293) 10 yrs. Or less 175 mg/daily
[22] Longitudinal • 75–174 mg caffeine Above or below • 3MS • women with above-median
(n = 1608) 6467 only • TICMS caffeine intake had a lower
(Driscoll et al., epidemiological 65–80 yrs FFQ at baseline median caffeine
• 175–189 mg caffeine female subjects • dementia risk of developing dementia
2016) study intake = 7.2 and
(n = 1784) 6.9 yrs. resp questionnaire or any cognitive impairment
• ≥190 mg caffeine compared to those consuming
(n = 1737) below median levels
Self-reported • Neurological
Questionnaire exam • those who consumed >200
[23] Longitudinal • <50 mg caffeine 587 cognitively • MMSE mg/day of caffeine had a
90–103 yrs. at enrolment and
(Paganini-Hill epidemiological • 50–199 mg caffeine normal at 36 months. • informant lower risk of dementia than
Mean = 93 ± 2.6 Leisure World
et at., 2016) study • 200+ mg caffeine baseline subjects questionnaire those who consumed
Cohort health
survey (1981–1985) • DQ <50 mg/day of caffeine
• CASI-short
• incidence of dementia was

• Incidence of inversely associated to the
• never coffee (n = 2048) 13,137 dementia
[24] Longitudinal consumption of coffee
• occasionally (n = 4194) non-cognitive reported to
(Sugiyama et al., epidemiological >65 yrs FFQ at baseline 5.7 yrs • The inverse relationship was
• 1–2 cup/day (n = 5246) disabled at the insurance
2016) study more remarkable among
• ≥3 cup/day (n = 1649) baseline database women, non-smokers, and
non-drinkers
• maintenance of working
memory behavioral
Self-reported performance in dCON
[25] Longitudinal 45 elderly sCON = 70.0 ± 4.3 • MMSE
• sCON, (n = 24) chronic coffee • reduced caffeine-induced
(Haller et al., epidemiological controls, 18 dCON = 73.4 ± 5.9 18 months
• WM task
• dCON, (n = 21) consumers brain activation changes in
2017) study with MCI MCI = 71.6 ± 4.7 in fMRI
• MCI, (n = 18) (1–3 cups/day) dCON compared to sCON.
• MR imaging
• caffeine is a cognitive
normalizer, not
cognitive enhancer
Molecules 2022, 27, 3737 14 of 41
Table 1. Cont.
• moderate coffee consumers
are less likely to be
• MR Imaging
categorized as dCON
•
[26] Longitudinal sCON = 73 ± 3 Substance • caffeine in sCON correlated
• sCON (n = 52) Neuropsychological
(Haller et al., epidemiological 145 subjects dCON = 74 ± 4 questionnaire at 3 yrs to fewer WM lesions and
• iCON (n = 62) assessments
2018) study iCON = 73 ± 3 baseline increased cerebral blood flow
• dCON (n = 32) • MSSE
but not in iCON and dCON
• HAD
• caffeine is a cognitive
• IADL
normalizer, not
cognitive enhancer
• <30 mg/day caffeine

• 30–111 mg/day 2475 cognitive
Willett • rate of cogitative preservation
[27] Longitudinal caffeine healthy female
semi-quantitative between the highest and
(Vercambre et al., epidemiological • 112–203 mg/day health 65+ yrs 5 yrs
food questionnaire • TICS lowest quantile of caffeine
2013) study caffeine professional with intake was equivalent to that
• 204–371 mg/day at baseline
CVD risk of 7 yrs. apart in age
caffeine
• >371 mg/day caffeine
• coffee drinkers at midlife had

• MMSE a markedly lower risk of
[18] Longitudinal • 0–2 cups/day (n = 223) Survey • DSM-IV developing dementia and AD
(Eskelinen et al., epidemiological • 3–5 cups/day (n = 641) 1409 individuals 65–79 yrs questionnaire at 21 yrs • NINCDS- • lowest risk of dementia and
2009) study • >5 cups/day (n = 542) baseline ADRDA AD was found among chronic
coffee consumers
(3–5 cups/day) at midlife
Molecules 2022, 27, 3737 15 of 41
Table 1. Cont.
• 0–1 unit/day
• women with >3 cups/day
(M = 27.4%, F = 24.6%) >65 yrs. Questions in the
7017 dementia showed less memory decline
[28] Longitudinal • 1–2 unit/day M = average standardized • BVRT
free subjects. Average = 3.4 ± than women consuming
(Ritchie et al., epidemiological (M = 32.4%, F = 31.5%) 73.6 ± 5.3 yrs. interview by • IST
Men(M) = 2820. 0.67 yrs ≤1 cup/day
2007) study • 2–3 unit/day F = average health professional • DSM-IV
Female(F) = 4197 • no relation between cognitive
(M = 27.0%, F = 27.5%) 73.8 ± 5.2 yrs at baseline • MMSE
decline and caffeine intake
• >3 unit/day
in men
(M = 13.2%, F = 16.4%)
• 0–115.5 mg/day
(n = 707)
3734 cognitive • no association between
• >115.5–188.0 mg/day
healthy Japanese midlife coffee intake and risk
[39] Longitudinal (n = 604) 24 h dietary recall
American men. 71–93 yrs. of cognitive impairment
(Gelber et al., epidemiological • >188.0–277.5 mg/day questionnaire at 25 yrs
Autopsy Mean = 52 yrs • CASI • higher caffeine intake is
2011) study (n = 784) entry (mid-life)
sub-group associated with a lower
• >277.5–415.0 mg/day
incidence of any type of brain
(n = 704) (n = 418)
lesions at autopsy
• >415.0–2673 mg/day
(n = 695)
• cross-sectionally reported
cognitive healthy higher caffeine intake was
Longitudinal subjects, 55% • MRI
associated with a lower
[40] epidemiological female. • LDST
• 0–1 cups/day
Mean = 59 ±
prevalence of lacunar infarcts,
(Araujo et al., study with cross-sectional FFQ at baseline 5 yrs • Stroop test
• >1–3 cups/day 7.2 yrs smaller hippocampus volume,
2016) cross-sectional (n = 2914), • WFT
• >3 cups/day and better
subgroup longitudinal • WLT
cognitive performance.
(n = 2454) • PBB
• These relationships are not
found longitudinally
Molecules 2022, 27, 3737 16 of 41
Table 1. Cont.
0–1 cups/
Questionnaire • Short follow-up, >3 cups/day,
Stratified • 0–1 cups/day (n = 360) Cognitively day = 70.3(8.6)
[41] baseline (n = 5408) • MMSE had a lower risk of dementia
Longitudinal • >1–3 cups/day healthy subjects >1–3 cups/ Mean = 13.2 ±
(Mirza et al., Follow-up • GMS than <1 cup/day
epidemiological (n = 1792) (0–4 yrs, n = 5408) day = 69.5(7.8) 5.4 yrs
2014) questionnaire • CAM-DEX • This relationship is not found
study • >3 cups/day (n = 3256) (>4 yrs. n = 4935) >3 cups/
(n = 4368) in long-follow-up
day = 66.3(7.3)
• Tea < 5×/year
(M = 26.5%, F = 21.1%)
• Tea 5–10×/year
(M = 11.7%, F = 10.6%)
• Tea 1–3×/month
(M = 18.6%, F = 18.4%)
• intake of coffee and tea
• Tea 1–4×/week
4809 cognitive modestly reduced rates of
(M = 21.8%, F = 22.6%)
healthy subjects cognitive decline in some
[29] Longitudinal • Tea ≥ 5×/week
Men(M) women
(M = 21.3%, F = 27.2) • MMSE
(Arab et al., epidemiological >65 yrs FFQ at baseline Median 7.9 yrs • no dose-effect relationship
• coffee < 5×/year (n = 2077)
2011) study among the women
(M = 30.8%, F = 37.4%) Women(W)
(n = 2722) • no relationship between
• coffee 5–10×/year
caffeine and cognition
(M = 6.7%, F = 6.2%)
among men
• coffee 1–3×/month
(M = 7.4%, F = 6.5%)
• coffee 1–4×/week
(M = 10.9%, F = 7.6%)
• coffee ≥ 5×/week
(M = 44.2%, F = 42.2%)
Avg =
• APOE e4 carrier 81.2 ± 3.4 yrs 8-item cognitive
[42] Longitudinal 2622 dementia- • no association between coffee
(n = 551) Carrier = 80.9 ± health food • CERAD
(Fischer et al., epidemiological free 10 yrs consumption and the
• APOE e4 non-carrier 3.4 yrs questionnaire at • SIDAM
2018) study participants incidence of dementia/AD
(n = 2071) Non-carrier = baseline
81.3 ± 3.4 yrs
Molecules 2022, 27, 3737 17 of 41
Table 1. Cont.
648 subjects
≥ 75 yrs. • caffeine intake was correlated
• Men followed up recruited
[30] Longitudinal Men = with reduced cognitive
(n = 128) Followed up
(Santos et al., epidemiological 70(67–73) FFQ at baseline 2–9 yrs • MMSE decline among women.
• Women followed up (n = 309)
2010) study Women = • no correlation found
(n = 181) Not followed up
71(68–74.5) among men
(n = 339)
1193 cognitive
[43] Cross-sectional Caffeine • no statistically significant
• Men (n = 473) healthy subjects • Aβ levels
(Ritchie et al., epidemiological ≥65 yrs questionnaire at Nill association between caffeine
• Women (n = 720) with plasma • MMSE
2014) study baseline interview and depression or Aβ levels
AB levels
• professional (n = 94)
Cross-sectional • skilled manual 351 subjects born MONICA food • coffee no relationship
[44]
epidemiological (n = 180) in 1936 and sat 64 yrs frequency NIll • MMSE cognition once account
(Kyle et al., 2010)
study • unskilled manual the MHT questionnaire for cognition
(n = 77)
• higher coffee intake was

significantly associated with
<2 cups/day = lower Aβ positivity
Cross-sectional Coffee intake • PET scan
[31] • <2 cups/day (n = 269) 411 adults 71.06 ± 7.73 yrs. • coffee intake was
epidemiological questions in Nill • MRI scan
(Kim et al., 2019) • ≥2 cups/day (n = 142) without dementia ≥2 cups/day not-associated with
study interview • CERAD-K
= 69.67 ± 8.43 yrs hypometabolism,
AD-signature region, and
WMH volume
>60 yrs. • an incidental finding that

• Dementia patient (D), 102 subjects
[32] Cross-sectional (C) = 68.9 ± Lifestyle increased coffee intake
(n = 52) without statins • MMSE
(Al-khateeb epidemiological 7.11 yrs. questionnaire at Nill demonstrated a 6.25-fold
• Healthy Control (C), use or substance • CDT
et al., 2014) study (D) = 70.7± baseline lower risk for
(n = 50) abuse history
7.63 yrs cognitive decline
Molecules 2022, 27, 3737 18 of 41
Table 1. Cont.
• episodic
memory • higher caffeine intake was
Young group = • executive related to better overall
Cross-sectional 634 cognitively function cognition
[33] • young group (n = 317) 64–71.5
epidemiological healthy T2D FFQ baseline Nill • semantic cate- • effect on cognition was
(West et al., 2019) • older group (n = 321) Older group =
study patients gorization6 amplified in the older group
71.5–84
• working (above median) compared to
memory the younger
• MR imaging
• tea/coffee
• coffee intake significantly
• None/day 493,944 subjects • PM decreased reaction time, pairs
[45] Cross-sectional • <1 cup/day without
Touchscreen • Pairs matching, Trail making test B,
(Cornelis et al., epidemiological • 1 cup/day self-reported 35–73 yrs Nill
questionnaire • FI and symbol digit substitution
2020) study • 2–3 cups/day neurological • RT • No relationship was
• 4–5 cups/day disease • SDS identified between cognitive
• 6–7 cups/day
function and CMS
• ≥8 cups/day
• Recent caffeine, whites

NO (n = 401,650)
• Recent caffeine, whites
434,900 subjects
• Among white and non-white
[46] Cross-sectional YES (n = 8533) without • PM participant, recent coffee
• Recent caffeine, Touchscreen • Pairs consumption was correlated
(Cornelis et al., epidemiological self-reported 35–73 yrs Nill
non-whites NO questionnaire • FI with higher RT performance
2020) study neurological
(n = 24,152)
disease
• RT but worse FI, Pairs, and
• Recent caffeine, PM performance
non-whites YES
(n = 565)
Molecules 2022, 27, 3737 19 of 41
Table 1. Cont.
• Caffeine intake was
[34] Cross-sectional • Q1 caffeine intake significant associated with
• Q2 caffeine intake ≥65 yrs 24-hr dietary • CERAD improved CERAD word
(Iranpour et al., epidemiological 1440 subjects Nill
• Q3 caffeine intake Mean = 69.14 yrs recall survey • DSST recall test
2020) study
• Q4 caffeine intake • this trend was enhanced
among men
• mean log transformed

WML/cranial volume ratio
[35] Cross-sectional Caffeine intake was lower for female chronic
(Ritchie et al., epidemiological • Men (n = 317) 641 subjects ≥65 yrs questions in Nill coffee consumer
• MR imaging
2010) study • Women (n = 324) interview • relationship did not extend
to men
• increased cerebral perfusion
in chronic coffee consumers
Cross-sectional • MFDF diet showed a lower

[47] • MFDF (n = 589) 765 cognitive • MMSE-KC
epidemiological ≥60 yrs FFQ Nill risk of cognitive impairment
(Kim et al., 2015) • WNC (n = 176) healthy subjects • CERAD
study compared to the western diet
• Vegetable and
non-processed diet
• ‘Traditional Australian
• Coffee, high sugar, high fat
[48] Cross-sectional diet,’ 65–90 yrs. Lifetime diet
352 cognitive diet had worse cognitive
(Hosking et al., epidemiological • ‘non-traditional Mean 73.12 questionnaire Nill
healthy subjects
• MMSE performance compared to the
2014) study Australian diet’ (SD = 5.47) yrs. (LDQ) ‘vegetable and
• ‘Coffee high-fat
non-processed diet’
• sugar extras diet
• Processed, high fat
• sugar extras diet
• 0 g/day (n = 710) • 0 g/day (n = 710)
• 1 to <266.4 g/day • 1 to <266.4 g/day • CERAD • CERAD
Cross- Cross- • Caffeinated coffee was associated•withCaffeinated coffee
[36] (n = 602) [36] (n = 602) • DSST • DSST
sectional sectional 2 × 4 h dietary recall improved
2 × 4 h dietary recall cognition improved cognitio
(Dong et al., • 266.4 (Dong = 2513 subjects•
to <495et(nal., ≥266.4
60 yrs • yrsAnimal
to <495 (n = 2513 subjectsNill ≥ 60 Nill • Animal
epidemiologic epidemiologic interview • •
No association between decaffeinated
interview No association bet
2020)
Molecules 2022, 27,al
3737 607) 2020) 607) fluency fluency 20 of 41
study al study coffee and cognition. coffee and cogniti
• ≥ 495 g/day (n = • ≥ 495 g/day (n = test test
594) 594)
Table 1. Cont.
Study Study Design >3 × 150 mg/day • mg/day
>3 × 150 Higher caffeine intake is associated
Cognition • with
Main Outcomes Higher caffeine in
• caffeinePopulation
treatment Groups • Sizecaffeine treatment
Age (yrs.) Caffeine Follow Up TestsGMV in the medial temporal
[37] [37] caffeine tablets reduced
caffeine tablets reduced GMV in t
Randomized (n = 10) 20 healthy
Randomized (n = 10) 20 healthy • MRI • Caffeinated
MRI
(Lin et al., • (Lin et al.,
0 g/day (n = 710) 18–35 yrs >Sweat test- caffeine 5.5 h18–35 yrs >Sweat test-lobe caffeine 5.5 h • coffee was
lobe
control study • placebo treatment subjects
control study • placebo treatment subjects • ECG • CERAD • ECG with
associated
2021) [36] Cross-sectional • 1 to <266.4
2021) g/day metabolite 2 × 4 h dietary • •Sleep
metabolite pattern not affected
DSST •
by caffeine
improved cognitionSleep pattern not a
(Dong et al., epidemiological(n = 10) (n = 602) (n = 10)
2513 subjects ≥60 yrs Nill
measurement recall interview •treatment
measurement Animal • No association between
treatment
2020) study • 266.4 to <495 (n = 607)
fluency test decaffeinated coffee
• ≥495 g/day (n = 594)
and cognition
>3 × 150 mg/day • Higher caffeine intake is

• caffeine treatment caffeine tablets
[37] Randomized 20 healthy associated with reduced GMV
(n = 10) 18–35 yrs >Sweat test- 5.5 h • MRI
in the medial temporal lobe
(Lin et al., 2021) control study • placebo treatment subjects
caffeine metabolite
• •acute caffeine administration
ECG
• •
induced a
acute caffeine adm
Sleep pattern not affected by
(n = 10) prefrontal activation incaffeine
HC and a more
measurement treatment prefrontal activati
HC = 68.3 HC = 68.3
34 subjects 34 subjects • MMSE • activation
diffuse posteromedial MMSE in MCI diffuse posterome
[38] [38] ± 2.8 yrs >Caffeine tablets ± 2.8 yrs >Caffeine tablets • acute caffeine administration
Randomized • 17 HC without •
Randomized 17 HC without • CDR • • induced
posterior displacement of working
CDR • posterior displace
a prefrontal
(Haller et al., (Haller et al., MCI = (200 mg) 30 min MCI = (200 mg) 30 min
control study • 17 MCI study •
neurological/ps
control 17 MCI neurological/ps • HAD •
memory-r after caffeine activation in HC
administrations
HAD and a more
memory-r after ca
2014) 2014) 70.7 ± 4.6 >placebo tablets 70.7 ± 4.6 >placebo tablets diffuse posteromedial
ychiatric history
34 subjects HC = 68.3 ± ychiatric history • LIDA •in MCI
MMSErepresents •
a compensatory
LIDA
activation in MCI in MCI represents
[38] yrs >Caffeine tablets yrs
Randomized • 17 HC without neurolog- 2.8 yrs •mechanism
CDR to counterbalance
• frontal mechanism
posterior displacement of to cou
(Haller et al., (200 mg) 30 min
control study • 17 MCI ical/psychiatric MCI = • HAD working memory-r after
2014)
history 70.7 ± 4.6 yrs
>placebo tablets
•
lobeLIDA
dysfunction lobeindysfunction
caffeine administrations
MCI represents a
compensatory mechanism to
The study did not find a neuroprotective caffeine linkdid
The study ; the
notstudy
find afound a neuroprotective
neuroprotective caffeinecaffeine link:
link ; the . found a neuroprotective caffeine
study link: .
counterbalance frontal
lobe dysfunction
The study did not find a neuroprotective caffeine link ; the study found a neuroprotective caffeine link: .
Molecules 2022, 27, 3737 21 of 41
4. In Vitro and In Vivo Studies

Due to unavoidable confounding in clinical studies, animal models have been used to
study the relationship between caffeine consumption on AD and cognition while exploring
possible action mechanisms. In this review, we identified and included 21 in vivo and
in vitro studies that directly explored the effect of caffeine in relation to AD. Twenty of
these studies showed caffeine was neuroprotective, and one determined that caffeine has
no effect on AD. We used this data to elaborate on the possible mechanism of action of the
neuroprotective effect of caffeine. Numerous studies examined other active components in
coffee that might be neuroprotective against AD, but these were excluded as this is outside
the scope of this review.
4.1. Effecting Membrane

Gastaldo et al., investigated the interaction of resveratrol, caffeine, β-carotene, and
epigallocatechin gallate (EGCG) on peptide aggregate by using synthetic membranes that
contained cross-β sheets of the membrane active fragment Aβ25–35 [49]. The effect on the size
and volume fraction of Aβ fragments was noted using microscopy (optical and fluoroscopy),
X-ray diffraction, UV-vis spectroscopy, and molecular dynamic simulations [49]. Gastaldo
et al., found that caffeine was membrane-active and simultaneously partitioned into the
synthetic membrane, where caffeine caused membrane thickening (38.3 Å ± 0.2) and
reduced membrane fluidity but did not affect the volume fraction of peptide aggregates [49].
The study also reported that caffeine attracted water and promoted the expulsion of plaques
from the membrane leading to more pronounced amyloid fibrils confirmed by microscopy
and x-ray diffraction [49]. The exact role this effect of caffeine plays in neuroprotective
AD is unclear, but membranes are reported to play a crucial role in the early stages of
peptide aggregation, where they provide a stable base for the crosslinking of neighboring
Aβ monomers thus by, caffeine causing early expulsion of peptides it prevents crosslinking
with neighboring monomers [49].
4.2. Altering APP Processing

Janitschke et al., studied the affinity of natural methylxanthines (MTX) (caffeine, theo-
bromine, theophylline) and synthetic MTX (propentofylline, pentoxifylline) on amyloid
precursor protein (APP) processing; also, further explored the molecular mechanism of
caffeine in the protective effect in AD [50]. They used human neuroblastoma (SH-SY5Y
WT and SH-SY5Y APP695 ), which was incubated in MTX (0.1 nmol per 1 µg protein) for
30 min before α and β-secretase activity measurements using Western blots (WB) [50].
Caffeine increased α-secreted APP to (131 ± 9.5%) and decreased direct β-secretase activity
(93.6 ± 1.0%), decreased BAEC1 gene expression (80.8 ± 6.8%) in SH-SY5Y APP695 [50].
Caffeine also increased direct α-secretase activity to (112.3 ± 3.0%) and increased ADAM10
protein levels (229.3 ± 22.4%) in SH-SY5Y WT [50]. Indicating that caffeine decreased total
secreted Aβ (by 15.5%) levels through elevation of non-amyloidogenic α-secretase APP
processing [50]. Furthermore, the study also reported that caffeine reduced reactive oxida-
tive species (ROS) to (48.3 ± 1.6%), reduced cholesterol levels (82.8 ± 3.7%), and reduced
Aβ aggregation too (46.7 ± 9.6%) [50]. A major strength of this study is that it explores the
molecular mechanism of reduction in Aβ levels in the hippocampus reported in caffeine-
treated APPswedish transgenic mice [51]. However, as an ex vivo research, the study does
not account for BBB and liver metabolism, limiting the results generalizability [50].
Arendash et al., investigated the long-term effect of chronic caffeine administration
(1.5 mg/day in drinking water, 4–9 months of age) on transgenic APPswedish mice (Tg) [51].
During the last eight weeks of the study, the mice were subjected to behavioral assess-
ment (open-field task, balance beam task, string-suspension task, Y-maze task, elevated
plus-maze, Morris water maze, circular platform task, platform recognition task, radial
arm water maze) and the rodents’ post-mortem brains were also analyzed to assess solu-
ble/insoluble Aβ levels, PCS1, and BACE levels, adenosine receptor density by WB, and
measure of brain adenosine levels [51]. The study also examined the effect of caffeine
Molecules 2022, 27, 3737 22 of 41
(0–10 µM) on Aβ production in vivo in APPswedish mice and Aβ generalization in N2a

neuronal cultures [51]. The Tg mice with chronic caffeine administration performed signifi-
cantly better than control Tg mice and, like WT mice, across multiple cognitive domains
(spatial learning/reference memory, working memory, recognition/identification) [51].
Post-mortem analysis revealed that caffeine-treated Tg mice had lower hippocampal Aβ-
levels, reduced presenilin 1/γ-secretase (PS1) and β-secretase (BACE1) levels, restored
brain adenosine levels, and unchanged A1 and A2A receptor density compared to control
Tg mice [51]. In vitro APPswedish mice and N2a neuronal cultures showed a caffeine
concentration-dependent decrease of Aβ production (Aβ1–40 and Aβ1–42 ) [51]. This study
suggests that the neuroprotective effect of caffeine is likely due to decreased Aβ production
achieved through modulation of BASE activity [51]. This study was the first to examine
the long-term effect of caffeine administration on WT and Tg mice [51]. However, further
research is required to explain the mechanisms involved in the modulation of BACE1 and
PS1 levels and activity [51].
In another study, Arendash et al., examined the effect of caffeine administration
(0.3 mg/mL) on aged transgenic APPswedish mice (18–19 months) showing impaired
working memory; after 4–5 weeks of caffeine treatment, the rodents were subjected to
behavioral testing (open-field task, balance beam task, string-suspension task, Y-maze
task, elevated plus-maze, Morris water maze, circular platform task, platform recognition
task, radial arm water maze) and the rodents’ post-mortem brains were also analyzed
(immunohistochemistry, Aβ ELISA, pcRaf-1 and PKA analysis) [52]. This study also in-
cluded a second experiment where 9-month-old Tg mice were gavaged with caffeine
(1.5 mg/twice daily for two weeks), after which the mice were sacrificed and subjected
to pcRaf-1 and PKA analysis [52]. In the third experiment carried out part of this study,
5.5-month WT mice were put on caffeine (0.3 mg/L) and at the age of 15–16 months were
subjected to 6-week behavioral screening [52]. Finally, the study also looked at the effect
of concentration-dependent caffeine administration (0–20 µM for 1h) or time-dependent
caffeine administration (20 µM for 0–180 min) on APPswedish mice N2a neuronal cul-
tures [52]. Arendash et al., found that caffeine administration on aging Tg mice with
impaired cognition showed markedly improved working memory and overall cognition
than Tg control mice (p < 0.05), who showed continued deterioration [52]. Caffeinated Tg
mice had lower Aβ deposition in the hippocampus (−40%) and entorhinal cortex (−46%)
and reduced soluble Aβ levels than Tg control [52]. Mechanistically they found that BACE1
suppression in Tg caffeinated involves cRaf-1/NFηB pathway (−27%, significantly lower
than in TG controls) and PKA (+25%, significantly higher than TG control) and that the
physiological concentration of caffeine (1–2 cups) was sufficient to reduce glycogen syn-
thases kinase 3 levels in N2a cells [52]. They also reported no cognitive benefit of long-term
caffeine treatment in WT mice [52]. The strength of this study is that it is the first to show
that caffeine can reverse AD cognitive impairment and proposes a mechanism of action for
the reduction of BAES1 [52].
Cao et al., studied the effect of acute (1.5 mg caffeine IP or gavage) and chronic caffeine
administration (2× daily 1.5 mg caffeine gavage for 7 days) on Aβ levels (plasma, CSF,
deposition) in transgenic APPswedish mice (Tg) and non-transgenic mice (NT) [53]. The
study also carried out in vivo microdialysis of living rodent hippocampus to study the
effect of acute caffeine administration on interstitial fluid Aβ levels of the hippocampus
by analyzing CSF and collected blood samples [53]. All the plasma samples underwent
neurochemical assessments (Aβ levels and cytokine expression profiles), and the chronic
caffeine group was also subjected radial arm water maze cognitive test [53]. Cao et al., acute
caffeine administration in Tg mice led to reducing Aβ levels in brain interstitial fluid and
plasma without affecting Aβ clearance [53]. Chronic caffeine administration led to reduced
plasma Aβ, and decreased soluble and deposited Aβ hippocampus and cortex [53]. Plasma
Aβ or caffeine levels did not correlate with brain Aβ levels or cognitive performance [53].
However, higher plasma caffeine was linked to reduced hippocampal neuroinflammatory
markers [53].
Molecules 2022, 27, 3737 23 of 41
4.3. Altering Excitation and Inhibition

A study investigated the long-term effect of early-life exposure to caffeine in THY-
Tau22 transgenic mice, an AD Tau pathology model [54]. A caffeine dose of 3 g/L (=humans
4 cup/day) was given to (THY-Tau22 and WT mice) caffeine group, starting 2 weeks before
mating to postnatal day 15, after which the offspring were given pure water, and the
water group (THY-Tau22 and WT mice) was never exposed to caffeine [54]. The mice’s
deficit in learning was accessed at 8 and 12 months using the Barnes maze test, in vitro
electrophysiology assessment of hippocampal CA1 pyramidal cells, then the tissue was
harvested for further biochemical and molecular evaluation [54]. The THY-Tau22 mice
caffeinated offspring developed cognitive deficits (spatial memory and learning) earlier at
eight months than water treated offspring at 12 months [54]. WT mice showed no difference
between caffeinated and water groups, unlike in Silva et al., but this is thought to be be-
cause object location memory was not tested in this study [54,74]. There was no correlation
between cardinal PTau, neuroinflammatory markers, and memory deficit in caffeinated off-
spring [54]. In vitro electrophysiology assessment showed that early life caffeine exposure
altered how glutamatergic and GABAergic circuits were affected by Tau pathology [54].
At eight months, caffeinated Tau mice had lower glutamatergic and GABAergic neuron
function, while at 12 months, their excitatory drives were decreased. Still, inhibitory drives
were increased compared to water Tau mice which had higher glutamatergic and GABAer-
gic drives compared to water WT mice at 8 and 12 months [54]. Thus, indicating a more
complex non-linear Tau-age-caffeine interaction than the predicted simple caffeine-induced
aging-like increase in glutamatergic and GABAergic drives of Tau mice [54]. The limitation
of this study is that only 2-time points of electrophysiology assessment were conducted,
which is insufficient to describe caffeine-induced changes in glutamatergic and GABAergic
drives of Tau mice; more in vivo electrophysical measurements are warranted to study this
hypothesis further [54].
4.4. Altering Protein Aggregation

Mancini et al., examined the ability of six compounds in coffee (caffeine, chlorogenic
acid, quinic acid, caffeic acid, quercetin, and phenylindole) at 25 mM to inhibit fibrilization
of Aβ and Tau or a-synuclein using thioflavin T (ThT) and thioflavin S (ThS) fluorescence
assay [55]. All instant coffee (light, dark and roast) inhibited Aβ and Tau aggregation
at 100 µg/mL [55]. However, caffeine had no effect on Aβ, Tau, and a-synuclein with
no measurable IC50 values [55]. This study’s limitation is that higher concertation of
caffeine that might inhibit aggregation was not tested [55]. However, other coffee com-
pounds are more potent inhibitors of protein aggregation, such as phenylindanes (inhibited
both Aβ and Tau fibrilization and Aβ oligomerization), and decaffeinated and caffeinated
coffee exhibit similar levels of inhibition of protein aggregation [55].
Laurent et al., investigated the effect of chronic caffeine intake (0.3 g/L drinking
water) on THY-Tau22 transgenic mouse’s progression of Tau pathology [56]. The rodents
were subjected to a Morris water maze cognitive test, biochemical analysis, mRNA
extraction, and caffeine metabolite sampling (brain and plasma) [56]. Chronic caffeine-
exposed (32.30 ± 2.21%) mice performed significantly better than control transgenic
mice (25.56 ± 2.57%) in spatial memory tests and were comparable to WT water ad-
ministered mice (34.04 ± 3.37%) [56]. Furthermore, the study reported that caffeine
administration to WT mice did not improve its cognitive performance [56]. Caffeinated
THY-Tau22 mice (−22.6 ± 7.0%) also had significantly lower Tau phosphor-isotopes
than THY-Tau22 control mice [56]. Caffeinated THY-Tau22 mice also had significantly
lower pro-inflammatory protein and oxidative stress markers than control Tau mice [56].
Importantly, this is the first study to demonstrate the neuroprotective effect of caffeine
against Tau pathology [56].
Molecules 2022, 27, 3737 24 of 41
4.5. Antioxidant Properties

Alzoubi et al., looked at the ability of caffeine (0.3 g/mL added to drinking water) to
reduce the cognitive decline caused by increased oxidative stress due to administration of L-
methionine (1.7 g/kg/day orally) for a treatment period of 4 weeks [57]. Then radial arm water
maze (RAWM) was used to measure cognition (spatial learning and memory), and a calori-
metric immunoassay was used to measure hippocampal tissue antioxidant biomarker [57].
Alzoubi et al., reported that L-methionine administration caused (short and long) term mem-
ory impairment (p < 0.05) while caffeine negated that effect [57]. L-methionine administration
caused reduced catalase and glutathione peroxidase (GPx) enzyme activities; reduced glu-
tathione (GSH) oxidized glutathione (GSSG) ratio compared to controls, while caffeine ad-
ministration normalized these effects [57]. A strength of this study is that it directly examined
the effect of caffeine on the hippocampus antioxidative protection system (GSH/GSSG ratio,
catalyze, GPx), allowing for a causal relationship to be determined [57].
4.6. Effect on BNDF Levels

Additionally, two studies also showed through in vivo experiments that caffeine was
able to prevent or reversed the reduction in brain-derived neurotrophic factor (BNDF) in
AD mice and mice on a high-fat diet (Table 2) [58,59]. This could be a key mechanism in
understanding the neuroprotective effect of caffeine in AD.
4.7. AR Antagonist Properties

Zhao et al., investigated if administration of 3 g/L caffeine (non-selective antagonist of
A2A R) in drinking water or gene knockout (A2A R KO mouse model) can elevate cognitive
impairment by reducing Tau-hyperphosphorylation induced by traumatic brain injury
(TBI) using a mouse model of moderate cortical impact [60]. The mice’s cognition was
assessed using the Morris water maze test (day 7 and week 4), and post-mortem (im-
munofluorescence, immunohistochemistry, Golgi staining, western blot) analysis was also
performed [60]. TBI-induced PTau was confirmed by increased Ser404 close to the site of
injury in the dentate gyrus of the contralateral hippocampus (24 h, 7 days, 4 weeks post-TBI)
and spatial memory impairment in Morris test (7 days and 4 weeks) [60]. Chronic Caffeine
treatment (starting 3-week before TBI) prevented TBI-induced PTau and spatial memory
deficit (7 days and 4 weeks) [60]. The study proposes a novel post-TBI (TBI-common
among AD patients 20–30%) mechanism of A2A R activation that triggers Tau hyperphos-
phorylation, causing memory impairment which may be normalized by chronic caffeine
administration [60]. Further research is required to explore the underlying mechanism for
the protective effect of A2A R on certain types of memory and why hyperphosphorylated
Tau appears in specific brain regions [60].
Bortolotto et al., investigated the effects of acute of caffeine (10 mg/kg, non-selective
AR antagonist), ZM241385 (10 µg/kg, A2A R antagonist), DPCPX (0.5 mg/kg, A1A R antago-
nist), dipyridamole (5 mg/kg, nucleoside transporter inhibitor), ELINA (100 µg/kg, adeno-
sine deaminase) on scopolamine (200 µM) induced memory loss in adult WT Zebrafish [61].
Then the fish was subjected to a battery of behavioral tests: inhibitory avoidance task,
exploratory assessment, and social interaction test [61]. Interestingly, caffeine pre-treatment
prevented scopolamine-induced amnesia (p < 0.0001 diff between training and test) in
the inhibitory avoidance test [61]. Caffeine administration also did not cause a significant
change in social interaction or exploratory assessment [61]. Other substances tested also
showed neuroprotective effects against scopolamine-induced memory deficits showing
that adenosine blockage can stop scopolamine-induced amnesia [61]. Caffeine’s ability to
prevent scopolamine-induced amnesia has been previously explored in rodent studies [61].
Li et al., studied if caffeine, through A3 R action, can reduce Aβ precursor protein
( AβPP) and LDL internalization, therefore reducing Aβ generation [62]. This study used
rat embryonic primary cerebral cortical neurons and human blastoma SH-SY5Y (express-
ing WT Aβ). [62]. All the cells were then treated with LDL (1 µg/mL) to measure LDL
internalization (known AD risk factor) [62]. Then the cells were selectively exposed to
Molecules 2022, 27, 3737 25 of 41
the following treatment: caffeine (200µM), selective AR-1a,2a,2b,3r antagonists, A3 R gene

knockout treatment. Then the cells were observed: immunoblotting, surface immunostain-
ing, RT-PCR, Lactate dehydrogenase (LDII- cell injury marker) [62]. Li et al., reported that
caffeine, A3 R antagonist, A3 R gene knockout showed a concentration-dependent reduction
in LDL internalization, suppression of LDL induced Aβ generation, and suppressed AβPP
internalization [62]. Interestingly caffeine suppressed LDL-induced Aβ1–40 & 1–42, but A3 R
gene knockout only suppressed Aβ1–40 [62]. Therefore, this study proposes a novel mecha-
nism where caffeine protects against LDL-enhanced AβPP internalization and processing
into Aβ by A3 R antagonism [62].
Espinosa et al., studied the effect of caffeine consumption (30 µM plasma) in adult
Wistar rats with sporadic AD (induced by streptozotocin (STZ), 5 µL) and its effect on
memory impairment, neuronal damage, A2A R hippocampal density [63]. After treatments,
the rodents were subjected to cognitive tests (behavioral analysis, novel object recognition
task), immunohistochemistry, immunoblotting, and quantitative-PCR [63]. Espinosa et al.,
reported that caffeine treatment prevented memory decline induced by STZ (F(1,38) = 46.195,
p < 0.001) but had no effect on controls [63]. Caffeine treatment was also shown to pre-
vent neurodegeneration induced by STZ in NeuN immunohistochemistry (F(1,8) = 12.49,
p < 0.0077), WBs also showed that caffeine prevented STZ induced increased A2A R hip-
pocampal density (F(1,20) = 4.83, p < 0.0399) but had no effect on controls [63]. This study
demonstrates that caffeine can prevent STZ-induced memory decline while simultaneously
controlling the hippocampal A2A R population, but the underlying mechanism for this
neuroprotective effect was not explored [63].
Dall’lgna et al., used CF1 adult mice with cognitive decline induced by Aβ injection
(3 nmol) and treated them with caffeine: acute, subchronic, prolonged, or combined [64].
Prolonged treatment consisted of (1 mg/mL of caffeine in drinking water for 12 days before
Aβ administration on day 7), sub-chronic caffeine administration (30 mg/kg caffeine for
2 days before and one day after Aβ administration), acute caffeine treatment intraperitoneal
(80 mg/kg caffeine, 30 min before Aβ administration), combined prolonged and acute
(1 mg/mL of caffeine in drinking water for 12 days and 80 mg/kg caffeine, 30 min before Aβ
administration) [64]. The same protocol of subchronic caffeine administration was followed
for selective A2A R antagonist (SCH58261, 0.5 mg/kg dose) [64]. Aβ administration caused
impaired performance in inhibitory avoidance and spontaneous alteration [64]. Acute
caffeine administration alone did not stop the Aβ-induced impaired performance, but
prolonged caffeine or selective A2A R treatment using (subchronic, chronic, and combined
prolonged) protocols showed a protective effect against cognitive decline [64].
4.8. Effect on Endolysosomes Dysfunction

Soliman et al., treated SH-SY5Y (over-expresses AβPP) with HIV-1 transactivator of
transcription (Tat) (200 µM) for two days in the presence/absence of caffeine (200 µM) [65].
Then they used quantified Aβ levels, vacuolar-ATPase protein, and phosphorylated Tau
levels [65]. Soliman et al., reported that HIV-1 Tat significantly increased levels of secreted
and intracellular levels of Aβ as well as hyperphosphorylated Tau [65]. HIV-1 Tat also
significantly reduced levels of vacuolar-ATPase, a major pathway that Endolysosomes use
to maintain an acidic environment [65]. Treatment with caffeine prevented this HIV-1 Tat-
induced increase in Aβ and Tau levels and prevented a decrease in vacuolar-ATPase [65].
The study hypothesized that caffeine could reduce HIV-1 Tat-induced lysosomal dysfunc-
tion, which allows the Endolysosomes to secrete H+ and maintain a lower PH. This helps
prevent an increase in Aβ and Tau levels [65]. Important to note that this study used 200 µM
of caffeine in the upper range of free caffeine plasma concentration, far higher than the
1–10 µM free caffeine plasma concentration seen after one cup of coffee [65].
4.9. Acetylcholinesterase Inhibition

Additionally, two studies showed that caffeine could inhibit acetylcholinesterase
(AChE) through in vitro experiments and computer modeling, which might be an important
Molecules 2022, 27, 3737 26 of 41
pathway for the neuroprotective effect of caffeine. However, these studies also reported
the plasma caffeine concentration needed for this effect is higher than normal plasma
concentration after consumption of coffee or energy drink [66,67].
4.10. Effect on Granulocyte-Colony Stimulating Factor, IL-6, and IL-10

Cao et al., examined the acute effects of decaffeinated coffee, caffeinated coffee (1.5 mg
caffeine), and pure caffeine (1.5 mg caffeine) on plasma cytokines measured (pre-treatment
and 30 min post-treatment) of 8-month-old transgenic APPswedish mice (Tg) and 8-month-
old non-transgenic mice (NTg) [68]. This study also examined the chronic effects of de-
caffeinated coffee, caffeinated coffee (0.75 mg caffeine), saline, and pure caffeine (0.75 mg
caffeine) administered twice weekly through gavage on plasma cytokines measured (on
the 13th month) of 10-month-old Tg and 10-month-old NTg [68]. Rodents were subjected
to cognitive interference task (chronic study only), and blood samples were analyzed using
Luminex assay (12 cytokines and chemokines), ELISA (plasma cons of theophylline, caf-
feine, and Aβ levels) [68]. The study reported that in the acute treatment, plasma levels
of a granulocyte-colony stimulating factor (GCSF), IL-6, and IL-10 were elevated for Tg
and NT mice treated with caffeinated coffee, not for caffeine treatment or decaffeinated
coffee treatment [68]. The chronic experiment identified that both caffeinated coffee and
caffeine treatment allowed better working memory preservation than Tg controls and to
the same level as NT controls [68]. This experiment also identified higher plasma GCSF
levels correlated with better cognition [68]. The mechanism of GCSF neuroprotection is hy-
pothesized to be through the recruitment of microglia from bone marrow, synaptogenesis,
and neurogenesis [68]. The strength of this study is that it demonstrated the importance of
the source of caffeine and how caffeine from coffee might have additional benefits due to
synergetic effects with an unknown component [68].
4.11. Effect on Aβ–Clearance

Qosa et al., investigated the potential of caffeine and rifampicin to enhance Aβ clear-
ance (induced by 30 nM of 125 I-Aβ40 ) across BBB in C57BL/6 WT mice [69]. For the
in vivo experiment, they treated WT mice with caffeine/rifampicin (20 mg/kg intraperi-
toneally, 2-week caffeine, and 3-week rifampicin) and subjected them to a brain efflux
index study [69]. Then for the In vitro experiment, they treated mouse brain endothe-
lial cells (bEnd3) with 50 µM of caffeine/rifampicin and analyzed them using PT-PCR
and western blotting [69]. Qosa et al., reported the brains of mice treated with caffeine
(BEI% = 80.4 ± 4.3%, p < 0.01) showed significantly higher Aβ clearance across BBB
compared to controls (BEI% = 62.4 ± 6.1%, p < 0.01) [69]. The in vitro studies also demon-
strated that caffeine treatment significantly upregulates the expression of P-glycoprotein
(P-GP), which is thought to be part of the mechanism that increases Aβ across BBB [69]. The
importance of this study is that it clearly demonstrates that the beneficial effects of caffeine
can be extended to clearance across BBB and identified the presence of another unidenti-
fied transport/receptor protein that acts in the same direction of low-density lipoprotein
receptor (LRPI) [69].
4.12. Studies Reporting no Effect of Caffeine

Shukitt-Hale et al., examined the effect of caffeine and coffee on cognition in WT Fisher
mice [70]. The mice were subjected for 8 weeks to varying diets of coffee (0, 0.165, 0.275,
0.55, 0.825) % of coffee extract in study 1 and for study 2 they used (0.387, 0.55)% coffee and
(0.0181, 0.0258)% caffeine [70]. The rodents were then subjected to a battery of psychomotor
(rod walking, wire suspension, plank walking, inclined screen, accelerating rod), cognitive
(Morris water maze), caffeine, and hydroxycinnamic acid concentration sampling [70]. The
study reported that 0.55% and 0.165% of coffee proved optimal for working and reference
memory [70]. However, in study 2, which compared the effect of pure caffeine, it was found
that it did not fully account for the protective effect of coffee, indicating that there might be
other players to the neuroprotective effect of coffee [70].
Molecules 2022, 27, 3737 27 of 41
Table 2. In vitro and in vivo studies summary.
Mechanism of
Study In-Vivo/In Vitro Study Study Methodology Main Outcomes
Neuroprotective Effect
â caffeine was membrane-active and
â Studied interaction of resveratrol,
simultaneously partitioned into the synthetic
caffeine, carotene, and epigallocatechin
membrane, where caffeine caused
gallate (EGCG) on Aβ peptide aggregate
membrane thickening
by using synthetic membranes that
[49] â caffeine attracted water and promoted the
In Vitro Effecting membrane contained cross-sheets of Aβ 25–35
Gastaldo et al., 2020 expulsion of plaques from the membrane
â The effect on the size and volume fraction
leading to more pronounced amyloid fibrils
of Aβ fragments noted using microscopy,
â caffeine by causing early expulsion of peptides
x-ray diffraction, UV-vis spectroscopy,
prevents crosslinking with neighboring
and molecular dynamic simulations
monomers and reduces peptide aggregation
â human neuroblastoma (SH-SY5Y WT and â caffeine decreased total secreted Aβ levels by

SH-SY5Y APP695) was incubated in MTX 15.5% through elevation of non-amyloidogenic
[50] (0.1 nmol per 1 g protein) 30 min before α
In Vitro Altering APP processing α-secretase APP processing
Janitschke et al., 2019 and β-secretase activity measurements â caffeine reduced ROS, cholesterol levels, and
using WB Aβ aggregation
â studied the effect of chronic caffeine

â Tg mice with chronic caffeine administration
administration (1.5 mg/day in drinking
performed significantly better than control Tg
water, 4–9 months of age) on transgenic
mice across multiple cognitive domains
APPswedish mice (Tg)
â caffeine treated Tg mice had lower
â last 8 weeks of the study, the mice were
In Vivo hippocampal Aβ levels, reduced presenilin 1
subjected to behavioral assessment
(PS1) and β-secretase (BACE1) levels, restored
[51] â rodent’s brain subjected to post-mortem
Altering APP processing brain adenosine levels, and unchanged A1 and
Arendash et al., 2006 WB analysis to measure soluble/insoluble
A2A receptor density compared to control
Aβ levels, PCS1, BACE and adenosine
Tg mice
levels, and adenosine receptor density
â APPswedish mice and N2a neuronal cultures

â Studied effect of caffeine (0–10 µM) on Aβ
In Vitro showed a caffeine concentration-dependent
production in vivo in APPswedish mice
decrease in Aβ1–40 and Aβ1–42 production
Molecules 2022, 27, 3737 28 of 41
Table 2. Cont.
Mechanism of
â effect of caffeine administration
(0.3 mg/mL) on aged transgenic
APPswedish mice (18–19 months)
showing impaired working memory
â after 4–5 weeks of caffeine treatment, the
mice were subjected to behavioral testing â caffeine administration on aging Tg mice
â post-mortem tissue was subjected to showed markedly improved working memory
immunohistochemistry, Aβ ELISA, and overall cognition than Tg control mice
pcRaf-1, and PKA analysis (p < 0.05)
â caffeinated Tg mice had lower Aβ deposition
In Vivo
â 9-month-old Tg mice were gavage with and lowered soluble Aβ levels than Tg control
[52] caffeine (1.5 mg/twice daily for 2 weeks), â mechanistically the neuroprotective effect of
Arendash et al., 2009 Altering APP processing â sacrificed and subjected to pcRaf-1 and caffeine involves BACE1 suppression in Tg
PKA analysis caffeinated through cRaf-1/NFηB pathway
and PKA
â 5.5-month WT mice were put on caffeine
(0.3 mg/L)
â age of 15–16 months was subjected to
6-week behavioral screening
â effect of concentration-dependent caffeine

administration (0–20 µM for 1h) or
â there was reduce glycogen synthases kinase 3
In Vitro time-dependent caffeine administration
levels in N2a cells
(20 µM for 0–180 min) on APPswedish
mice N2a neuronal cultures
Molecules 2022, 27, 3737 29 of 41
Table 2. Cont.
Mechanism of
â acute (1.5 mg caffeine IP or gavage) and
â acute and chronic caffeine administration in Tg
chronic caffeine administration (2× daily
mice led to reduced Aβ levels in brain
1.5 mg caffeine gavage for 7 days) on Aβ
interstitial fluid and plasma
levels of (Tg) and (NT)
[53]
In Vivo Altering APP processing
Cao et al., 2009
â microdialysis of living rodent
hippocampus to study the effect of acute â plasma Aβ or caffeine levels did not correlate
caffeine administration on interstitial with brain Aβ levels or cognitive performance
fluid Aβ levels
â investigated the long-term effect of

early-life exposure to caffeine in
THY-Tau22 transgenic mice
â caffeine dose of 3 g/L was given to â in vitro electrophysiology assessment showed
parental THY-Tau22 Mice and WT mice, that early life caffeine exposure altered
starting 2 weeks before mating and glutamatergic and GABAergic circuits
[54] Altering excitation
In Vivo continued to postnatal day 15 â complex non-linear Tau-age-caffeine
Zappettini et al., 2019 and inhibition
â then learning of offspring Tg and WT interaction rather than the predicted simple
mice was accessed at 8 and 12 months caffeine-induced aging-like increase in
â offspring Tg and WT mice were subjected glutamatergic and GABAergic drives
to in vivo electrophysiology examination
of hippocampal neuronal activity and
post-mortem biochemical analysis
â studied ability of caffeine, chlorogenic

acid, quinic acid, caffeic acid, quercetin,
[55] â caffeine on its own could not interfere with Aβ,
In Vitro Altering protein aggregation and phenylindole at 25 mM to inhibit
Mancini et al., 2018 Tau aggregation, and Aβ oligomerization
fibrilization of Aβ and Tau
â using (ThT) and (ThS) fluorescence assay
Molecules 2022, 27, 3737 30 of 41
Table 2. Cont.
Mechanism of
â effect of chronic caffeine intake (0.3 g/L â chronic caffeine Tg mice performed
drinking water) on THY-Tau22 mouse significantly better than control Tg mice
[56]
In Vivo Altering protein aggregation â rodent subjected to cognitive test, â Caffeinated Tg mice had significantly lower
Laurent et al., 2014
biochemical analysis, mRNA extraction, Tau phosphor-isotopes, pro-inflammatory and
and caffeine metabolite sampling oxidative stress markers than Tg control mice
â caffeine (0.3 g/mL added to drinking â L-methionine administration caused (short and
water) to reduce the cognitive decline long) term memory impairment while caffeine
caused by increased oxidative stress due negated that effect
[57] to administration of L-methionine (1.7 â L-methionine administration caused reduced
In Vivo Antioxidant properties
Alzoubi et al., 2018 g/kg/day orally) for a treatment period catalyze and GPx enzyme activities; reduced
of 4 weeks GSH, GSSG ratio compared to controls, while
â cognition and hippocampal tissue caffeine administration normalized
antioxidant markers were assessed these effects
â effect of caffeine on rats placed on a

high-fat diet
â Rodents’ hippocampus was subjected to â caffeine treatment was sufficient to prevent
microdialysis and then spontaneous high-fat diet weight gain and high-fat diet
[58] alternating testing to test the working memory impairment
In Vivo Effect on BNDF levels
Moy et al., 2013 memory of rodents. â caffeine diet prevented reduction in BNDF
â Post mortem the rodent brains were induced by a high-fat diet and allowed
subjected to histology, WB, and maintenance of synaptic plasticity
enzyme-linked immunosorbent assay
(BNDF quantification)
Molecules 2022, 27, 3737 31 of 41
Table 2. Cont.
Mechanism of
â effect of caffeine 0.75 mg/day or
â caffeine administration Tg mice significantly
1.5 mg/day on saline vehicle treatment
improved cognitive performance compared to
[59] for 8 weeks on the expression of BNDF
In Vivo Effect on BNDF levels control Tg mice
Han et al., 2013 and TrKB receptors in Tg mice
â dose-response increase of hippocampal BNDF
â rodents subjected to a Morris water maze
and TrKB expression in caffeinated Tg mice
test on and WB
â administration of 3 g/L caffeine in

drinking water or A2A R KO mouse
model can increase cognitive impairment
by reducing Tau-hyperphosphorylation
â post-TBI mechanism of A2A R activation that
induced by TBI mouse model
[60] triggers hyperphosphorylation of Tau, causing
In Vivo AR antagonist properties â cognition was assessed using the Morris
Zhao et al., 2017 memory impairment may be normalized by
water maze test (day 7 and
chronic caffeine administration
week 4 post-treatment)
â post-mortem (immunohistochemistry,
Golgi staining, Western blotting were
also performed
â effects of acute of caffeine (10 mg/kg),

ZM241385 (10 µg/kg,), DPCPX (0.5
mg/kg), dipyridamole (5 mg/kg), ELINA
(100 µg/kg,) on scopolamine (200 µM)
[61] â caffeine pre-treatment prevented
In Vivo AR antagonist properties induced memory loss in adult
Bortolotto et al., 2015 scopolamine-induced amnesia
WT Zebrafish
â subjected to behavioral tests such as
inhibitory avoidance task, exploratory
assessment, and social interaction test
Molecules 2022, 27, 3737 32 of 41
Table 2. Cont.
Mechanism of
â caffeine (200 µM), selective
AR-1a,2a,2b,3r antagonists, A3 R gene
knockout treatment can reduce AβPP and
â caffeine, A3 R antagonist, A3 R gene knockout
LDL internalization, therefore reducing
showed a concentration-dependent reduction
[62] Aβ generation in rat embryonic primary
In Vitro AR antagonist properties in LDL internalization, suppression of
Li et al., 2015 cerebral cortical neurons and human
LDL-induced Aβ generation, suppressed
blastoma SH-SY5Y
AβPP internalization
â the cells were examined by Western
blotting, surface immunostaining,
RT-PCR, Lactate dehydrogenase
â effect of caffeine consumption (30 µm

plasma) in adult Wistar rats with
â caffeine can prevent STZ-induced memory
[63] sporadic AD (induced by STZ, 5 µL)
In Vivo AR antagonist properties decline while simultaneously controlling the
Espinosa et al., 2013 â rodents were subjected to cognitive tests,
hippocampal A2A R population
immunohistochemistry, immunoblotting,
and quantitative-PCR
â CF1 adult mice with cognitive decline

induced by Aβ injection (3 nmol)and
â prolonged caffeine or selective A2A R treatment
treated them with caffeine or selective
[64] using: (subchronic, chronic, and combined
In-Vivo AR antagonist properties A2A R treatment (acute, subchronic,
Dall’Igna et al., 2007 prolonged) protocols showed a protective
prolonged, or combined)
effect against cognitive decline
â Subjected to inhibitory avoidance and
spontaneous alteration cognitive tests
â SH-SY5Y with HIV-1 Tat (200 µM) for

2 days in the presence/absence of â caffeine was able to prevent HIV-1 Tat induced
[65] Effect on
In Vitro caffeine (200 µM) increase in Aβ and Tau levels and prevent a
Soliman et al., 2017 endolysosomes dysfunction
â quantified Aβ levels, vacuolar-ATPase, decrease in vacuolar-ATPase
and phosphorylated Tau protein levels
Molecules 2022, 27, 3737 33 of 41
Table 2. Cont.
Mechanism of
â effect of xanthine (caffeine, pentoxifylline,
[66] In Vitro Acetylcholinesterase propentofylline) on the inhibition of
â caffeine was a weak AChE inhibitor
Mohamed et al., 2013 In Silico inhibition AChE through in vitro and molecular
modeling studies
â used standard Elman test and in silico

â caffeine is a strong non-competitive inhibitor of
[67] In vitro Acetylcholinesterase examinations to determine whether
AChE and a weak non-competitive inhibitor of
Pohanka et al., 2013 In silico inhibition caffeine could inhibit human BChE
BChE
and ACh
â examined the effects of decaffeinated

coffee, caffeinated coffee (1.5 mg caffeine),
and pure caffeine (1.5 mg caffeine) on
plasma cytokines measured in â acute treatment, plasma levels of GCSF, IL-6,
8-month-old Tg and NTg mice IL-10 was elevated for Tg and NTg mice
â chronic effects of decaffeinated coffee, treated with caffeinated coffee only
Effect on granulocyte-colony caffeinated coffee (0.75 mg caffeine), â chronic experiment, it was identified that both
[68]
In Vivo stimulating factor, IL-6, saline, and pure caffeine (0.75 mg caffeinated coffee and caffeine treatment
Cao et al., 2011
and IL-10 caffeine) administered twice weekly allowed better preservation of working
through gavage on plasma cytokines of memory compared to NTg controls and higher
10-month-old Tg and NTg plasma GCSF levels correlated with
â Rodents were subjected to cognitive better cognition
interference task, and blood samples were
analyzed using Luminex assay
and ELISA
gavage on plasma cytokines of 10-month-old Tgplasma
gavage on and cytokines of 10-month-old Tg and
NTg NTg
  interference
Rodents were subjected to cognitive task,
Rodents were subjected to cognitive interference task,
Molecules 2022, 27, 3737 and blood samples were analyzed using
andLuminex
blood samples were analyzed using Luminex 34 of 41
assay and ELISA assay and ELISA
 
treated WT mice with caffeine/rifampicin (20WT
treated mg/kg
mice with caffeine/rifampicin (20 mg/kg
Table 2. Cont. intraperitoneally, 2-week caffeine, and 3-week
intraperitoneally, 2-weekbrains of mice
caffeine, and treated
3-week  showed
with caffeine significa
brains of mice t
rifampicin) and subjected them to a brain efflux index
rifampicin) and higher
subjected themA𝛽toclearance
a brain acrossindex
efflux BBB compared
highertoA𝛽
control
clear
[69] [69]Mechanism of
Study In-Vivo/In In vivo
Vitro Study Effect on In vivo study Effect
Study on
Methodology study Main Outcomes
Qosa et al., Qosa et al.,
In vitro 𝐴𝛽– clearance In vitro 𝐴𝛽– clearance
2012 2012  â treated mouse 50 μMof treated mouse bEnd3 with 50 μM of
treated WTbEnd3 with caffeine/rifampicin
mice with
 
in vitro caffeine treatment significantly in upregulates
vitro caffeinet
caffeine/rifampicin and analyzed 2-week
(20 mg/kg intraperitoneally, them using PT-PCR
caffeine/rifampicin
â and analyzed
brains them using
of mice treated PT-PCRshowed
with caffeine
expression of P-GP-mechanism that increases
expressionA𝛽ofacr
P
caffeine, and 3-week rifampicin) and
and WB and WB significantly higher Aβ clearance across BBB
subjected them to a brain efflux compared to controls
[69] In Vivo Effect on index study
Qosa et al., 2012 In Vitro Aβ-clearance  WT mice were subjected for 8 weeks  to WT varying dietssubjected for 8 weeks to varying diets
mice were
âof coffee extract (0%, 0.165%, 0.275%,
treated mouse bEnd3 with 50 µM of 0.55%,
of coffee0.825%)
extract
â (0%, 0.165%,
in vitro 0.275%,
caffeine 0.55%,significantly
treatment 0.825%)
in study 1
caffeine/rifampicin and analyzed them in study 1 upregulates the expression of P-GP-mechanism
 studyusing
2 WTPT-PCR
mice were and subjected
WB to coffee
study(0.387%,
2 WT mice that wereincreases
subjectedAβtoacross
coffeeBBB
(0.387%,
[70] [70]
0.55%) and (0.0181%, 0.0258%) for 8 weeks 0.55%) and (0.0181%,  0.0258%)
pure caffeine
for 8 did
weeks  forpure
not fully account the protective
caffeine di
Shukitt-hale In vivo Shukitt-hale Nill In vivo â Nill
 then WT mice were
the rodents subjected
were for
subjected to8 weeks
thento
psychomotor and were coffee
the rodents subjected to psychomotor and coffee
et al., 2013 et al., 2013 varying diets of coffee extract (0%,
cognitive testing cognitive testing
 0.165%,
the brain and0.275%, 0.55%, 0.825%)
serum concentrations  of inthe
study
caffeine
brain1 and serum concentrations of caffeine and
â study 2 WT mice were subjected
hydroxycinnamic acid metabolites were to coffee
also
hydroxycinnamic acid metabolites were also
(0.387%, 0.55%) and (0.0181%, 0.0258%)
[70] recorded recorded â pure caffeine did not fully account for the
In Vivo Nill for 8 weeks
Shukitt-hale et al., 2013 protective effect of coffee
â then the rodents were subjected to
Study did not find a neuroprotective Study
caffeine link:
did not; find
psychomotorstudy found
aand a neuroprotective
neuroprotective
cognitive caffeine
caffeine link:
testing link:found
; study . a neuroprotective caffeine link: .
â the brain and serum concentrations of
caffeine and hydroxycinnamic acid
metabolites were also recorded
Study did not find a neuroprotective caffeine link: ; study found a neuroprotective caffeine link: .
Molecules 2022, 27, 3737 35 of 41
5. Discussion
When considering the results of these studies, variances in caffeine metabolism
between individuals that lead to varying plasma caffeine concentration must be con-
sidered as it can cause variation in physiological effects. 99% of the caffeine from
beverages, i.e., tea and coffee, is rapidly absorbed through the gastrointestinal tract
and distributed through body water, bypassing the liver (V = 0.7 L/kg), reaching peak
plasma concentration within 15–120 min [75,76]. The main route of caffeine metabolism
for clearance in humans (70–80%) is through the N-3-demethylation to paraxanthine
pathway, which CYP1A2 carries out in the liver [77]. The existence of multiple variants
of CYP1A2 causes variability in caffeine metabolism and plasma caffeine concentration
in individuals [78]. Furthermore, smokers’ caffeine metabolism is also accelerated,
leading to lower plasma caffeine concentration [79]. Other than that, exogenous estro-
gen has also been shown to inhibit caffeine metabolism and increase plasma caffeine
concentration [79].
This review identified a few studies that examined the neuroprotective effect of
coffee/caffeine by using brain imaging (MRI, fMRI) to assess the changes to brain struc-
ture and neural activation patterns [25,26,31,35,38,43]. Studies by Haller et al. [25,38]
showed in early cognitive decline, there is an increase in compensatory basal activity
diffused through the posto-temporal region of the brain, which increases the brain’s
sensitivity to the neuroprotective action of caffeine [25,38]. Furthermore, MRI studies
by Ritchie et al. [35] and Haller et al. [26] showed that caffeine reduces the amount
of white matter lesion/cranial volume in cognitively stable elders, contributing to
cognitive decline in Dementia/AD [26,43]. Ritchie et al. [43] also showed increased
cerebral perfusion in chronic coffee consumers, indicating a possible neuroprotective
mechanism of coffee [43]. Moreover, Gelber et al. [39] found high caffeine levels were
associated with a lower-odds of having any brain lesion types at autopsy [39]. How-
ever, an epidemiological study by Kim et al. [31] did not find any association between
coffee intake and hypometabolism, atrophy of AD signature, and WMH volume; in-
stead, it found that coffee exerted a neuroprotective effect by reducing the levels of
Aβ [31].
Current literature also seems to support the notion that caffeine/coffee acts as a cogni-
tive normalizer instead of a cognitive enhancer, and as such healthy adults or those with
deteriorating cognition receive little benefit from coffee/caffeine treatment [25,26,33,38].
West et al., found that elderly participants with mild cognitive decline showed higher sen-
sitivity to caffeine than healthy younger diabetics [33]. Furthermore, Haller et al., found no
changes in neural activation among healthy adults but increased posto-temporal activation
in those with MCI, further supporting the notion that caffeine does not act as a cognitive
enhancer [38]. Other than that, caffeine/coffee cannot significantly enhance the cognitive
function of those suffering from severe cognitive decline [25,26]. Haller et al., showed that
caffeine reduces the amount of white matter lesion/cranial volume and increases cerebral
perfusion in cognitively stable elders but did not extend the same benefits to elders with
deteriorating cognition [26]. Furthermore, Haler et al., in another study used fMRI to study
the neural activation induced by caffeine for participants with deteriorating cognition;
although this study showed that caffeine reduced cognitive decline in dCON, it did not
show the same level of caffeine-induced neural activation in them as seen in those with
sCON [25].
The literature also shows that the neuroprotective effect of caffeine/coffee can be
confounded by gender, but the evidence is not definitive for either gender, and further
research is needed [24,28,34,43]. Two studies (Ritchie et al. [28,43]) only found a statistically
significant neuroprotective effect of caffeine among women in the study population but
not males [28,43]. Furthermore, Sugiyama et al. [24] found an overall neuroprotective
effect of coffee, but this was enhanced among the female cohort and non-smokers and non-
drinkers [24]. However, Iranpour et al., in a crude model, found the neuroprotective effect
of caffeine extended to both genders, but after adjusting for confounding, found a weak
Molecules 2022, 27, x FOR PEER REVIEW 37 of 42
Molecules 2022, 27, 3737 36 of 41
[34]. The exact mechanism for the difference in neuroprotective properties of caffeine/cof-
feepositive
between correlation
gender is forunclear
the neuroprotective
and warrants effect of caffeine
further only among
research; the males
however, it hasonly
been[34].
hypoth-
The exact mechanism for the difference in neuroprotective properties of caffeine/coffee
esized that this may be due to differences in caffeine metabolism, pharmacodynamics, or
between gender is unclear and warrants further research; however, it has been hypothesized
hormonal
that this influence
may be due[34,28].
to differences in caffeine metabolism, pharmacodynamics, or hormonal
Although there is evidence from the clinical studies suggesting that caffeine con-
influence [28,34].
sumption is protective
Although there isagainst
evidence AD cognitive
from decline,
the clinical further
studies clinical
suggesting studies
that arecon-
caffeine required
sumption
to prove thisislink.
protective
Ideally,against AD cognitive
to examine decline,
this link, therefurther
would clinical
need studies
to be anare required
epidemiological
to prove
study with this
largelink. Ideally,
sample to examine
size, this link,surveys
with multiple there would need to
collecting be an epidemio-
extensive data on con-
logical study with large sample size, with multiple surveys collecting
founding variables to be adjusted, including data on CYP genotype. The study should extensive data also
on confounding variables to be adjusted, including data on CYP genotype. The study
use biological markers (blood tests) to assess caffeine to reduce recall bias, and caffeine
should also use biological markers (blood tests) to assess caffeine to reduce recall bias,
dataandshould
caffeinebedata
collected
shouldat bemultiple
collected at points (incl:
multiple midlife)
points (incl:during
midlife)extended follow-up to
during extended
assess
follow-up to assess changes in behavior. Furthermore, cognition should also be using
changes in behavior. Furthermore, cognition should also be measured mea- veri-
fiedsured
methods, i.e., MMSE,
using verified CDR,
methods, i.e.,CERAD,
MMSE, CDR, during follow-up.
CERAD, during A sub-group
follow-up. should also be
A sub-group
should also
randomly be randomly
chosen chosentoand
and subjected subjected
brain imaging to brain
at set imaging
intervalsatduring
set intervals during
follow-up to iden-
follow-up to identify changes in neural architecture before shifting
tify changes in neural architecture before shifting in behavior. This subgroup should alsoin behavior. This
subgroup should
be subjected also be subjected
to post-mortem analysis to post-mortem
to confirm the analysis to confirm
presence the presence
and stage and
of AD (Figure 1).
stage of AD (Figure 1).
Figure 1. Summary of findings.

Figure 1. Summary of findings.
This review also analyzed in vivo and in vitro studies that directly examined the re-
lationship between caffeine on AD and cognition, which shows strong evidence that caf-
feine is neuroprotective against AD through multiple mechanisms. These studies also sug-
gest possible mechanisms of caffeine’s neuroprotective effect. Four of these studies show
that caffeine alters APP processing to a non-amyloid pathway, reducing AD burden and
cognitive decline [50–53]. Furthermore, five of these studies show that caffeine‘s neuro-
Molecules 2022, 27, 3737 37 of 41
This review also analyzed in vivo and in vitro studies that directly examined the
relationship between caffeine on AD and cognition, which shows strong evidence that
caffeine is neuroprotective against AD through multiple mechanisms. These studies
also suggest possible mechanisms of caffeine’s neuroprotective effect. Four of these
studies show that caffeine alters APP processing to a non-amyloid pathway, reducing
AD burden and cognitive decline [50–53]. Furthermore, five of these studies show that
caffeine‘s neuroprotective effect is due to its ability as a non-selective adenosine receptor
antagonist [60–64]. Other than that, studies have also shown that caffeine’s neuropro-
tective effect is due to its ability to alter protein aggregation [55,56]. We also found
evidence that caffeine is able to reduce BNDF levels [58,59]. Caffeine also reduces acetyl-
cholinesterase activity, a mechanism for its neuroprotective effect against AD [66,67].
Some studies showed that the neuroprotective effect of caffeine is by: affecting mem-
brane properties [49], changing GABAergic and glutamatergic neurotransmission [54],
reducing endolysosome dysfunction [65], increasing GCSF function [68], and increasing
Aβ clearance [69].
Through this review, we also identified a few in vivo and in vitro studies that were
excluded because they did not directly examine the relationship between caffeine and
AD. These, however, also posited possible mechanisms of action for the neuroprotective
effect of caffeine. Reznikov et al., showed that caffeine significantly enhanced C-Fos
expression in the horizontal limb of the diagonal band of Broca [80]. This is thought to
explain why AD patients lose their olfactory sense first and that the cognitive enhancing
effect of caffeine may be through activation of the basal cholinergic forebrain [80]. Fur-
thermore, Vila-luna et al., showed that the caffeinated group of mice had greater fourth
and fifth-order basal dendrites branching in CA1 pyramidal neurons. Laurent et al.
showed caffeine’s ability through A2R antagonism/knock out for normalization of
hippocampal GSH/GSSG ratio, global reduction in Tau hyperphosphorylation, and
neuroinflammatory markers [81].
6. Conclusions
This review found suggestive evidence in clinical studies to propose that caffeine
is neuroprotective against dementia and possibly AD, but further studies are required
to strengthen this link (Figure 1). The clinical studies also point out that caffeine is a
cognitive normalizer and not a cognitive enhancer. Although clinical studies show that the
neuroprotective effect of caffeine may be confounded by gender, it is not conclusive. The
review also found strong evidence based on in vivo and in vitro studies that caffeine has
some positive effects in AD models, but further studies are warranted to identify all the
mechanistic pathways of the neuroprotective effect of caffeine in AD.
Author Contributions: Conceptualization, Y.M.M.Y. and A.K.; methodology, Y.M.M.Y. and A.K.;
writing—original draft preparation, Y.M.M.Y. and A.K.; writing—review and editing, Y.M.M.Y.,
A.K., H.J.W. and R.L.M.F.; supervision, A.K., H.J.W. and R.L.M.F.; project administration, A.K.;
funding acquisition, A.K. and R.L.M.F. All authors have read and agreed to the published version of
the manuscript.
Funding: This work was supported by Alzheimer’s New Zealand (AK; 3718869), Freemasons New
Zealand (AK; 3719321 and 501029), Alzheimer’s New Zealand Charitable Trust (AK; 3720863), and
the Health Research Council of New Zealand (RF and HW; 3627373).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.
Molecules 2022, 27, 3737 38 of 41
References
1. Scheltens, P.; Blennow, K.; Breteler, M.M.B.; de Strooper, B.; Frisoni, G.B.; Salloway, S.; Van der Flier, W.M. Alzheimer’s disease.
Lancet 2016, 388, 505–517. [CrossRef]
2. Prince, M.J.; Wimo, A.; Guerchet, M.M.; Ali, G.C.; Wu, Y.-T.; Prina, M. World Alzheimer Report 2015—The Global Impact of Dementia:
An Analysis of Prevalence, Incidence, Cost and Trends; Alzheimer’s Disease International: London, UK, 2015.
3. Karran, E.; Mercken, M.; Strooper, B.D. The amyloid cascade hypothesis for Alzheimer’s disease: An appraisal for the development
of therapeutics. Nat. Rev. Drug Discov. 2011, 10, 698–712. [CrossRef]
4. Karch, C.M.; Goate, A.M. Alzheimer’s Disease Risk Genes and Mechanisms of Disease Pathogenesis. Biol. Psychiatry 2015, 77,
43–51. [CrossRef]
5. Arnsten, A.F.T.; Datta, D.; Del Tredici, K.; Braak, H. Hypothesis: Tau pathology is an initiating factor in sporadic Alzheimer’s
disease. Alzheimer’s Dement. 2021, 17, 115–124. [CrossRef]
6. Genin, E.; Hannequin, D.; Wallon, D.; Sleegers, K.; Hiltunen, M.; Combarros, O.; Bullido, M.J.; Engelborghs, S.; De Deyn, P.;
Berr, C.; et al. APOE and Alzheimer disease: A major gene with semi-dominant inheritance. Mol. Psychiatry 2011, 16, 903–907.
[CrossRef]
7. Norton, S.; Matthews, F.E.; Barnes, D.E.; Yaffe, K.; Brayne, C. Potential for primary prevention of Alzheimer’s disease: An analysis
of population-based data. Lancet Neurol. 2014, 13, 788–794. [CrossRef]
8. De Bruijn, R.F.; Bos, M.J.; Portegies, M.L.; Hofman, A.; Franco, O.H.; Koudstaal, P.J.; Ikram, M.A. The potential for prevention of
dementia across two decades: The prospective, population-based Rotterdam Study. BMC Med. 2015, 13, 132. [CrossRef]
9. Panza, F.; Solfrizzi, V.; Barulli, M.R.; Bonfiglio, C.; Guerra, V.; Osella, A.; Seripa, D.; Sabba, C.; Pilotto, A.; Logroscino, G. Coffee,
tea, and caffeine consumption and prevention of late-life cognitive decline and dementia: A systematic review. J. Nutr. Health
Aging 2015, 19, 313–328. [CrossRef]
10. Kim, J.; Lee, K.W. Coffee and Its Active Compounds Are Neuroprotective; Elsevier: Amsterdam, The Netherlands, 2015; pp. 423–427.
11. Singh, S.S.; Rai, S.N.; Birla, H.; Zahra, W.; Kumar, G.; Gedda, M.R.; Tiwari, N.; Patnaik, R.; Singh, R.K.; Singh, S.P. Effect of
Chlorogenic Acid Supplementation in MPTP-Intoxicated Mouse. Front. Pharm. 2018, 9, 757. [CrossRef]
12. Balakrishnan, R.; Azam, S.; Cho, D.-Y.; Su-Kim, I.; Choi, D.-K. Natural Phytochemicals as Novel Therapeutic Strategies to Prevent
and Treat Parkinson’s Disease: Current Knowledge and Future Perspectives. Oxidative Med. Cell. Longev. 2021, 2021, 6680935.
[CrossRef]
13. Fahanik-Babaei, J.; Baluchnejadmojarad, T.; Nikbakht, F.; Roghani, M. Trigonelline protects hippocampus against intracerebral
Aβ(1–40) as a model of Alzheimer’s disease in the rat: Insights into underlying mechanisms. Metab. Brain Dis. 2019, 34, 191–201.
[CrossRef]
14. PubChem Compound Summary for CID 2519, Caffeine. Available online: https://pubchem.ncbi.nlm.nih.gov/compound/
Caffeine (accessed on 17 May 2022).
15. Yoshimura, H. The Potential of Caffeine for Functional Modification from Cortical Synapses to Neuron Networks in the Brain.
Curr. Neuropharmacol. 2005, 3, 309–316. [CrossRef]
16. Arendash, G.W.; Cao, C. Caffeine and Coffee as Therapeutics Against Alzheimer’s Disease. J. Alzheimer’s Dis. 2010, 20, S117–S126.
[CrossRef]
17. Berthou, F.; Ratanasavanh, D.; Alix, D.; Carlhant, D.; Riche, C.; Guillouzo, A. Caffeine and theophylline metabolism in newborn
and adult human hepatocytes; comparison with adult rat hepatocytes. Biochem. Pharm. 1988, 37, 3691–3700. [CrossRef]
18. Eskelinen, M.H.; Ngandu, T.; Tuomilehto, J.; Soininen, H.; Kivipelto, M. Midlife Coffee and Tea Drinking and the Risk of Late-Life
Dementia: A Population-Based CAIDE Study. J. Alzheimers Dis. 2009, 16, 85–91. [CrossRef]
19. Lelo, A.; Miners, J.O.; Robson, R.; Birkett, D.J. Assessment of caffeine exposure: Caffeine content of beverages, caffeine intake,
and plasma concentrations of methylxanthines. Clin. Pharmacol. Ther. 1986, 39, 54–59. [CrossRef]
20. Cao, C.; Loewensteine, D.A.; Lin, X.; Zhang, C.; Wang, L.; Duara, R.; Wu, Y.; Giannini, A.; Bai, G.; Cai, J.; et al. High Blood
Caffeine Levels in MCI Linked to Lack of Progression to Dementia. J. Alzheimers Dis. 2012, 30, 559–572. [CrossRef]
21. Solfrizzi, V.; Panza, F.; Imbimbo, B.P.; D’Introno, A.; Galluzzo, L.; Gandin, C.; Misciagna, G.; Guerra, V.; Osella, A.;
Baldereschi, M.; et al. Coffee Consumption Habits and the Risk of Mild Cognitive Impairment: The Italian Longitudinal Study on
Aging. J. Alzheimers Dis. 2015, 47, 889–899. [CrossRef]
22. Driscoll, I.; Shumaker, S.A.; Snively, B.M.; Margolis, K.L.; Manson, J.E.; Vitolins, M.Z.; Rossom, R.C.; Espeland, M.A. Relationships
Between Caffeine Intake and Risk for Probable Dementia or Global Cognitive Impairment: The Women’s Health Initiative
Memory Study. J. Gerontol. Ser. A-Biol. Sci. Med. Sci. 2016, 71, 1596–1602. [CrossRef]
23. Paganini-Hill, A.; Kawas, C.H.; Corrada, M.M. Lifestyle Factors and Dementia in the Oldest-old the 90 + Study. Alzheimer Dis.
Assoc. Disord. 2016, 30, 21–26. [CrossRef]
24. Sugiyama, K.; Tomata, Y.; Kaiho, Y.; Honkura, K.; Sugawara, Y.; Tsuji, I. Association between Coffee Consumption and Incident
Risk of Disabling Dementia in Elderly Japanese: The Ohsaki Cohort 2006 Study. J. Alzheimers Dis. 2016, 50, 491–500. [CrossRef]
[PubMed]
25. Haller, S.; Montandon, M.-L.; Rodriguez, C.; Moser, D.; Toma, S.; Hofmeister, J.; Giannakopoulos, P. Caffeine impact on working
memory-related network activation patterns in early stages of cognitive decline. Neuroradiology 2017, 59, 387–395. [CrossRef]
[PubMed]
Molecules 2022, 27, 3737 39 of 41
26. Haller, S.; Montandon, M.-L.; Rodriguez, C.; Herrmann, F.R.; Giannakopoulos, P. Impact of Coffee, Wine, and Chocolate
Consumption on Cognitive Outcome and MRI Parameters in Old Age. Nutrients 2018, 10, 1391. [CrossRef] [PubMed]
27. Vercambre, M.N.; Berr, C.; Ritchie, K.; Kang, J.H. Caffeine and cognitive decline in elderly women at high vascular risk. J.
Alzheimers Dis. 2013, 35, 413–421. [CrossRef] [PubMed]
28. Ritchie, K.; Carriere, I.; de Mendonca, A.; Portet, F.; Dartigues, J.F.; Rouaud, O.; Barberger-Gateau, P.; Ancelin, M.L. The
neuroprotective effects of caffeine—A prospective population study (the Three City Study). Neurology 2007, 69, 536–545.
[CrossRef] [PubMed]
29. Arab, L.; Biggs, M.L.; O’Meara, E.S.; Longstreth, W.T.; Crane, P.K.; Fitzpatrick, A.L. Gender differences in tea, coffee, and cognitive
decline in the elderly: The Cardiovascular Health Study. J. Alzheimers Dis. 2011, 27, 553–566. [CrossRef]
30. Santos, C.; Lunet, N.; Azevedo, A.; de Mendonça, A.; Ritchie, K.; Barros, H. Caffeine intake is associated with a lower risk of
cognitive decline: A cohort study from Portugal. J. Alzheimers Dis. 2010, 20 (Suppl. 1), S175–S185. [CrossRef]
31. Kim, J.W.; Byun, M.S.; Yi, D.; Lee, J.H.; Jeon, S.Y.; Jung, G.; Lee, H.N.; Sohn, B.K.; Lee, J.-Y.; Kim, Y.K.; et al. Coffee intake and
decreased amyloid pathology in human brain. Transl. Psychiatry 2019, 9, 27. [CrossRef]
32. Al-khateeb, E.; Al-zayadneh, E.; Al-dalahmah, O.; Alawadi, Z.; Khatib, F.; Naffa, R.; Shafagoj, Y. Relation between Copper, Lipid
Profile, and Cognition in Elderly Jordanians. J. Alzheimers Dis. 2014, 41, 203–211. [CrossRef]
33. West, R.K.; Ravona-Springer, R.; Livny, A.; Heymann, A.; Shahar, D.; Leroith, D.; Preiss, R.; Zukran, R.; Silverman, J.M.;
Schnaider, M. Age Modulates the Association of Caffeine Intake with Cognition and With Gray Matter in Elderly Diabetics. J.
Gerontol. Ser. A-Biol. Sci. Med. Sci. 2019, 74, 683–688. [CrossRef]
34. Iranpour, S.; Saadati, H.M.; Koohi, F.; Sabour, S. Association between caffeine intake and cognitive function in adults; effect
modification by sex: Data from National Health and Nutrition Examination Survey (NHANES) 2013–2014. Clin. Nutr. 2020, 39,
2158–2168. [CrossRef] [PubMed]
35. Ritchie, K.; Artero, S.; Portet, F.; Brickman, A.; Muraskin, J.; Beanino, E.; Ancelin, M.-L.; Carriere, I. Caffeine, Cognitive
Functioning, and White Matter Lesions in the Elderly: Establishing Causality from Epidemiological Evidence. J. Alzheimers Dis.
2010, 20, S161–S166. [CrossRef] [PubMed]
36. Dong, X.; Li, S.; Sun, J.; Li, Y.; Zhang, D. Association of Coffee, Decaffeinated Coffee and Caffeine Intake from Coffee with
Cognitive Performance in Older Adults: National Health and Nutrition Examination Survey (NHANES) 2011–2014. Nutrients
2020, 12, 840. [CrossRef] [PubMed]
37. Lin, Y.-S.; Weibel, J.; Landolt, H.-P.; Santini, F.; Meyer, M.; Brunmair, J.; Meier-Menches, S.M.; Gerner, C.; Borgwardt, S.;
Cajochen, C.; et al. Daily Caffeine Intake Induces Concentration-Dependent Medial Temporal Plasticity in Humans: A Multimodal
Double-Blind Randomized Controlled Trial. Cereb. Cortex 2021, 31, 3096–3106. [CrossRef]
38. Haller, S.; Montandon, M.-L.; Rodriguez, C.; Moser, D.; Toma, S.; Hofmeister, J.; Sinanaj, I.; Lovblad, K.-O.; Giannakopoulos, P.
Acute Caffeine Administration Effect on Brain Activation Patterns in Mild Cognitive Impairment. J. Alzheimers Dis. 2014, 41,
101–112. [CrossRef]
39. Gelber, R.P.; Petrovitch, H.; Masaki, K.H.; Ross, G.W.; White, L.R. Coffee Intake in Midlife and Risk of Dementia and its
Neuropathologic Correlates. J. Alzheimers Dis. 2011, 23, 607–615. [CrossRef]
40. Araujo, L.F.; Mirza, S.S.; Bos, D.; Niessen, W.J.; Barreto, S.M.; van der Lugt, A.; Vernooij, M.W.; Hofman, A.; Tiemeier, H.;
Ikram, M.A. Association of Coffee Consumption with MRI Markers and Cognitive Function: A Population-Based Study. J.
Alzheimers Dis. 2016, 53, 451–461. [CrossRef]
41. Mirza, S.S.; Tiemeier, H.; de Bruijn, R.F.A.G.; Hofman, A.; Franco, O.H.; Kiefte-de Jong, J.; Koudstaal, P.J.; Ikram, M.A. Coffee
consumption and incident dementia. Eur. J. Epidemiol. 2014, 29, 735–741. [CrossRef]
42. Fischer, K.; Melo van Lent, D.; Wolfsgruber, S.; Weinhold, L.; Kleineidam, L.; Bickel, H.; Scherer, M.; Eisele, M.; van den
Bussche, H.; Wiese, B.; et al. Prospective Associations between Single Foods, Alzheimer’s Dementia and Memory Decline in the
Elderly. Nutrients 2018, 10, 852. [CrossRef]
43. Ritchie, K.; Ancelin, M.L.; Amieva, H.; Rouaud, O.; Carriere, I. The association between caffeine and cognitive decline: Examining
alternative causal hypotheses. Int. Psychogeriatr. 2014, 26, 581–590. [CrossRef]
44. Kyle, J.; Fox, H.C.; Whalley, L.J. Caffeine, Cognition, and Socioeconomic Status. J. Alzheimers Dis. 2010, 20, S151–S159. [CrossRef]
[PubMed]
45. Cornelis, M.C.; Weintraub, S.; Morris, M.C. Caffeinated Coffee and Tea Consumption, Genetic Variation and Cognitive Function
in the UK Biobank. J. Nutr. 2020, 150, 2164–2174. [CrossRef] [PubMed]
46. Cornelis, M.C.; Weintraub, S.; Morris, M.C. Recent Caffeine Drinking Associates with Cognitive Function in the UK Biobank.
Nutrients 2020, 12, 1969. [CrossRef] [PubMed]
47. Kim, J.; Yu, A.; Choi, B.Y.; Nam, J.H.; Kim, M.K.; Oh, D.H.; Yang, Y.J. Dietary Patterns Derived by Cluster Analysis are Associated
with Cognitive Function among Korean Older Adults. Nutrients 2015, 7, 4154–4169. [CrossRef]
48. Hosking, D.E.; Nettelbeck, T.; Wilson, C.; Danthiir, V. Retrospective lifetime dietary patterns predict cognitive performance in
community-dwelling older Australians. Br. J. Nutr. 2014, 112, 228–237. [CrossRef]
49. Gastaldo, I.P.; Himbert, S.; Ram, U.; Rheinstadter, M.C. The Effects of Resveratrol, Caffeine, beta-Carotene, and Epigallocatechin
Gallate (EGCG) on Amyloid-beta(25-35) Aggregation in Synthetic Brain Membranes. Mol. Nutr. Food Res. 2020, 64, 2000632.
[CrossRef]
Molecules 2022, 27, 3737 40 of 41
50. Janitschke, D.; Nelke, C.; Lauer, A.A.; Regner, L.; Winkler, J.; Thiel, A.; Grimm, H.S.; Hartmann, T.; Grimm, M.O.W. Effect of
Caffeine and Other Methylxanthines on A beta-Homeostasis in SH-SY5Y Cells. Biomolecules 2019, 9, 689. [CrossRef]
51. Arendash, G.W.; Schleif, W.; Rezai-Zadeh, K.; Jackson, E.K.; Zacharia, L.C.; Cracchiolo, J.R.; Shippy, D.; Tan, J. Caffeine protects
Alzheimer’s mice against cognitive impairment and reduces brain β-amyloid production. Neuroscience 2006, 142, 941–952.
[CrossRef]
52. Arendash, G.W.; Mori, T.; Cao, C.; Mamcarz, M.; Runfeldt, M.; Dickson, A.; Rezai-Zadeh, K.; Tane, J.; Citron, B.A.; Lin, X.; et al.
Caffeine reverses cognitive impairment and decreases brain amyloid-beta levels in aged Alzheimer’s disease mice. J. Alzheimers
Dis. 2009, 17, 661–680. [CrossRef]
53. Cao, C.; Cirrito, J.R.; Lin, X.; Wang, L.; Verges, D.K.; Dickson, A.; Mamcarz, M.; Zhang, C.; Mori, T.; Arendash, G.W.; et al. Caffeine
suppresses amyloid-beta levels in plasma and brain of Alzheimer’s disease transgenic mice. J. Alzheimers Dis. 2009, 17, 681–697.
[CrossRef]
54. Zappettini, S.; Faivre, E.; Ghestem, A.; Carrier, S.; Buee, L.; Blum, D.; Esclapez, M.; Bernard, C. Caffeine Consumption during
Pregnancy Accelerates the Development of Cognitive Deficits in Offspring in a Model of Tauopathy. Front. Cell. Neurosci. 2019,
13, 438. [CrossRef] [PubMed]
55. Mancini, R.S.; Wang, Y.; Weaves, D.F. Phenylindanes in Brewed Coffee Inhibit Amyloid-Beta and Tau Aggregation. Front. Neurosci.
2018, 12, 735. [CrossRef] [PubMed]
56. Laurent, C.; Eddarkaoui, S.; Derisbourg, M.; Leboucher, A.; Demeyer, D.; Carrier, S.; Schneider, M.; Hamdane, M.; Müller, C.E.;
Buée, L.; et al. Beneficial effects of caffeine in a transgenic model of Alzheimer’s disease-like tau pathology. Neurobiol. Aging 2014,
35, 2079–2090. [CrossRef] [PubMed]
57. Alzoubi, K.H.; Mhaidat, N.M.; Obaid, E.A.; Khabour, O.F. Caffeine Prevents Memory Impairment Induced by Hyperhomocys-
teinemia. J. Mol. Neurosci. 2018, 66, 222–228. [CrossRef] [PubMed]
58. Moy, G.A.; McNay, E.C. Caffeine prevents weight gain and cognitive impairment caused by a high-fat diet while elevating
hippocampal BDNF. Physiol. Behav. 2013, 109, 69–74. [CrossRef] [PubMed]
59. Han, K.; Jia, N.; Li, J.; Yang, L.; Min, L.-Q. Chronic caffeine treatment reverses memory impairment and the expression of brain
BNDF and TrkB in the PS1/APP double transgenic mouse model of Alzheimer’s disease. Mol. Med. Rep. 2013, 8, 737–740.
[CrossRef]
60. Zhao, Z.A.; Zhao, Y.; Ning, Y.L.; Yang, N.; Peng, Y.; Li, P.; Chen, X.Y.; Liu, D.; Wang, H.; Chen, X.; et al. Adenosine A(2A)
receptor inactivation alleviates early-onset cognitive dysfunction after traumatic brain injury involving an inhibition of tau
hyperphosphorylation. Transl. Psychiatry 2017, 7, e1123. [CrossRef]
61. Bortolotto, J.W.; de Melo, G.M.; Cognato, G.d.P.; Moreira Vianna, M.R.; Bonan, C.D. Modulation of adenosine signaling prevents
scopolamine-induced cognitive impairment in zebrafish. Neurobiol. Learn. Mem. 2015, 118, 113–119. [CrossRef]
62. Li, S.; Geiger, N.H.; Soliman, M.L.; Hui, L.; Geiger, J.D.; Chen, X. Caffeine, Through Adenosine A(3) Receptor-Mediated Actions,
Suppresses Amyloid-beta Protein Precursor Internalization and Amyloid-beta Generation. J. Alzheimers Dis. 2015, 47, 73–83.
[CrossRef]
63. Espinosa, J.; Rocha, A.; Nunes, F.; Costa, M.S.; Schein, V.; Kazlauckas, V.; Kalinine, E.; Souza, D.O.; Cunha, R.A.; Porciuncula, L.O.
Caffeine Consumption Prevents Memory Impairment, Neuronal Damage, and Adenosine A(2A) Receptors Upregulation in the
Hippocampus of a Rat Model of Sporadic Dementia. J. Alzheimers Dis. 2013, 34, 509–518. [CrossRef]
64. Dall’Igna, O.P.; Fett, P.; Gomes, M.W.; Souza, D.O.; Cunha, R.A.; Lara, D.R. Caffeine and adenosine A(2a) receptor antagonists
prevent beta-amyloid (25-35)-induced cognitive deficits in mice. Exp. Neurol. 2007, 203, 241–245. [CrossRef] [PubMed]
65. Soliman, M.L.; Geiger, J.D.; Chen, X. Caffeine Blocks HIV-1 Tat-Induced Amyloid Beta Production and Tau Phosphorylation. J.
Neuroimmune Pharmacol. 2017, 12, 163–170. [CrossRef]
66. Mohamed, T.; Osman, W.; Tin, G.; Rao, P.P.N. Selective inhibition of human acetylcholinesterase by xanthine derivatives: In vitro
inhibition and molecular modeling investigations. Bioorganic Med. Chem. Lett. 2013, 23, 4336–4341. [CrossRef] [PubMed]
67. Pohanka, M.; Dobes, P. Caffeine Inhibits Acetylcholinesterase, But Not Butyrylcholinesterase. Int. J. Mol. Sci. 2013, 14, 9873–9882.
[CrossRef]
68. Cao, C.; Wang, L.; Lin, X.; Mamcarz, M.; Zhang, C.; Bai, G.; Nong, J.; Sussman, S.; Arendash, G. Caffeine synergizes with another
coffee component to increase plasmato cognitive benefits in Alzheimer’s mice. J. Alzheimers Dis. 2011, 25, 323–335. [CrossRef]
[PubMed]
69. Qosa, H.; Abuznait, A.H.; Hill, R.A.; Kaddoumi, A. Enhanced Brain Amyloid-β Clearance by Rifampicin and Caffeine as a
Possible Protective Mechanism Against Alzheimer’s Disease. J. Alzheimers Dis. 2012, 31, 151–165. [CrossRef]
70. Shukitt-Hale, B.; Miller, M.G.; Chu, Y.-F.; Lyle, B.J.; Joseph, J.A. Coffee, but not caffeine, has positive effects on cognition and
psychomotor behavior in aging. Age 2013, 35, 2183–2192. [CrossRef]
71. Culm-Merdek, K.E.; von Moltke, L.L.; Harmatz, J.S.; Greenblatt, D.J. Fluvoxamine impairs single-dose caffeine clearance without
altering caffeine pharmacodynamics. Br. J. Clin. Pharm. 2005, 60, 486–493. [CrossRef]
72. Fredholm, B.B.; Bättig, K.; Holmén, J.; Nehlig, A.; Zvartau, E.E. Actions of Caffeine in the Brain with Special Reference to Factors
That Contribute to Its Widespread Use. Pharmacol. Rev. 1999, 51, 83–133.
73. Alves, R.C.; Almeida, I.M.; Casal, S.; Oliveira, M.B. Isoflavones in coffee: Influence of species, roast degree, and brewing method.
J. Agric. Food Chem. 2010, 58, 3002–3007. [CrossRef]
Molecules 2022, 27, 3737 41 of 41
74. Silva, C.G.; Metin, C.; Fazeli, W.; Machado, N.J.; Darmopil, S.; Launay, P.S.; Ghestem, A.; Nesa, M.P.; Bassot, E.; Szabo, E.; et al.
Adenosine Receptor Antagonists Including Caffeine Alter Fetal Brain Development in Mice. Sci. Transl. Med. 2013, 5,
197ra104–197ra191. [CrossRef] [PubMed]
75. Bonati, M.; Latini, R.; Galletti, F.; Young, J.F.; Tognoni, G.; Garattini, S. Caffeine disposition after oral doses. Clin. Pharmacol. Ther.
1982, 32, 98–106. [CrossRef] [PubMed]
76. Arnaud, M.J. The pharmacology of caffeine. Prog. Drug Res. 1987, 31, 273–313. [CrossRef] [PubMed]
77. Begas, E.; Kouvaras, E.; Tsakalof, A.; Papakosta, S.; Asprodini, E.K. In vivo evaluation of CYP1A2, CYP2A6, NAT-2 and xanthine
oxidase activities in a Greek population sample by the RP-HPLC monitoring of caffeine metabolic ratios. Biomed. Chromatogr.
2007, 21, 190–200. [CrossRef] [PubMed]
78. Thorn, C.F.; Aklillu, E.; McDonagh, E.M.; Klein, T.E.; Altman, R.B. PharmGKB summary: Caffeine pathway. Pharm. Genom. 2012,
22, 389–395. [CrossRef] [PubMed]
79. Pollock, B.G.; Wylie, M.; Stack, J.A.; Sorisio, D.A.; Thompson, D.S.; Kirshner, M.A.; Folan, M.M.; Condifer, K.A. Inhibition of
caffeine metabolism by estrogen replacement therapy in postmenopausal women. J. Clin. Pharm. 1999, 39, 936–940. [CrossRef]
80. Reznikov, L.R.; Pasumarthi, R.K.; Fadel, J.R. Caffeine elicits c-Fos expression in horizontal diagonal band cholinergic neurons.
Neuroreport 2009, 20, 1609–1612. [CrossRef]
81. Laurent, C.; Burnouf, S.; Ferry, B.; Batalha, V.L.; Coelho, J.E.; Baqi, Y.; Malik, E.; Mariciniak, E.; Parrot, S.; Van der Jeugd, A.; et al.
A(2A) adenosine receptor deletion is protective in a mouse model of Tauopathy. Mol. Psychiatry 2016, 21, 97–107. [CrossRef]

Molecules 27 03737

Uploaded by

Copyright:

Available Formats

Molecules 27 03737

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Molecules 27 03737

Uploaded by

Copyright:

Available Formats

molecules

Molecules 2022, 27, 3737. https://doi.org/10.3390/molecules27123737 https://www.mdpi.com/journal/molecules

3. Clinical Studies Investigating the Relationship between Caffeine and Cognition

3.2. Cross-Sectional Studies

3.3. Randomized Control Studies

• habitual moderate coffee

• incidence of dementia was

• <30 mg/day caffeine

• coffee drinkers at midlife had

• higher coffee intake was

>60 yrs. • an incidental finding that

• Recent caffeine, whites

• mean log transformed

Cross-sectional • MFDF diet showed a lower

>3 × 150 mg/day • Higher caffeine intake is

4. In Vitro and In Vivo Studies

4.1. Effecting Membrane

4.2. Altering APP Processing

(0–10 µM) on Aβ production in vivo in APPswedish mice and Aβ generalization in N2a

4.3. Altering Excitation and Inhibition

4.4. Altering Protein Aggregation

4.5. Antioxidant Properties

4.6. Effect on BNDF Levels

4.7. AR Antagonist Properties

the following treatment: caffeine (200µM), selective AR-1a,2a,2b,3r antagonists, A3 R gene

4.8. Effect on Endolysosomes Dysfunction

4.9. Acetylcholinesterase Inhibition

4.10. Effect on Granulocyte-Colony Stimulating Factor, IL-6, and IL-10

4.11. Effect on Aβ–Clearance

4.12. Studies Reporting no Effect of Caffeine

Table 2. In vitro and in vivo studies summary.

â human neuroblastoma (SH-SY5Y WT and â caffeine decreased total secreted Aβ levels by

â studied the effect of chronic caffeine

â APPswedish mice and N2a neuronal cultures

â effect of concentration-dependent caffeine

â investigated the long-term effect of

â studied ability of caffeine, chlorogenic

â effect of caffeine on rats placed on a

â administration of 3 g/L caffeine in

â effects of acute of caffeine (10 mg/kg),

â effect of caffeine consumption (30 µm

â CF1 adult mice with cognitive decline

â SH-SY5Y with HIV-1 Tat (200 µM) for

â used standard Elman test and in silico

â examined the effects of decaffeinated

Figure 1. Summary of findings.

You might also like