Received: 3 February 2021 / Accepted: 4 August 2021 / Published online: 15 August 2021 © The Author(s) 2021

Purpose There is an ever-growing body of literature examining the relationship between dietary omega-3 polyunsaturated fatty acids (ω3 PUFAs) and cerebral structure and function throughout life. In light of this, the use of ω3 PUFAs, namely, long-chain (LC) ω3 PUFAs (i.e., eicosapentaenoic acid and docosahexaenoic acid), as a therapeutic strategy to mitigate cognitive impairment, and progression to Alzheimer’s disease is an attractive prospect. This review aims to summarise evidence reported by observational studies and clinical trials that investigated the role of LC ω3 PUFAs against cognition impairment and future risk of Alzheimer’s disease.

Methods Studies were identified in PubMed and Scopus using the search terms “omega-3 fatty acids”, “Alzheimer’s disease” and “cognition”, along with common variants. Inclusion criteria included observational or randomised controlled trials (RCTs) with all participants aged ≥ 50 years that reported on the association between LC ω3 PUFAs and cognitive function or biological markers indicative of cognitive function linked to Alzheimer’s disease.

Results Evidence from 33 studies suggests that dietary and supplemental LC ω3 PUFAs have a protective effect against cognitive impairment. Synaptic plasticity, neuronal membrane fluidity, neuroinflammation, and changes in expression of genes linked to cognitive decline have been identified as potential targets of LC ω3 PUFAs. The protective effects LC ω3 PUFAs on cognitive function and reduced risk of Alzheimer’s disease were supported by both observational studies and RCTs, with RCTs suggesting a more pronounced effect in individuals with early and mild cognitive impairment.

Conclusion The findings of this review suggest that individuals consuming higher amounts of LC ω3 PUFAs are less likely to develop cognitive impairment and that, as a preventative strategy against Alzheimer’s disease, it is most effective when dietary LC ω3 PUFAs are consumed prior to or in the early stages of cognitive decline.

Keywords Polyunsaturated fatty acids · Docosahexaenoic acid · Eicosapentaenoic acid · Alzheimer’s disease · Neurodegeneration · Cognitive impairment

Abbreviations 5-LOX 5-Lipoxygenase Aβ Amyloid-beta peptide AD Alzheimer’s disease ADAS-cog Alzheimer’s disease assessment scale cogni-

tive subscale ADL Activities of daily living ALA Alpha-linolenic acid AMT Abbreviated mental test ApoE Apolipoprotein E ARA Arachidonic acid

* Michael A. Zulyniak [email protected]

1 Nutritional Epidemiology Group, School of Food Science and Nutrition, University of Leeds, Leeds LS2 9JT, UK

CDR Clinical dementia rating CDT Clock-drawing test CIBIC Clinician’s interview-based impression of

change COWAT Controlled oral word association test COX Cyclooxygenase DHA Docosahexaenoic acid EPA Eicosapentaenoic acid HVLT-DR Hopkins verbal learning test with delayed

recall IADL Instrumental activities of daily living IL-6 Interleukin-6 LA Linoleic acid MCI Mild cognitive impairment MMSE Mini-mental state examination PPAR Peroxisome proliferator-activated receptor

Vol.:(0123456789)1 3

LC PUFAs Long-chain polyunsaturated fatty acids RAVLT Rey auditory verbal learning test RBANS Repeatable battery for assessment of neu-

ropsychological status sIL-1RII Soluble interleukin-1 receptor type-II TICS-M Modified telephone interview for cognitive

status TMT Trail making test TNF-α Tumour necrosis factor-alpha WAIS-III Wechsler adult intelligence scale third

edition WAIS-R Revised wechsler Adult intelligence scale WMS-R Revised wechsler memory scale ω3 PUFAs Omega-3 polyunsaturated fatty acids ω6 PUFAs Omega-6 polyunsaturated fatty acids

Encompassing a collection of syndromes, dementia is characterised by progressive impairment of memory, language, behaviour and visuospatial function, leading to compromised independence [1]. It is estimated to affect 50 million people globally [2] and is caused by a number of diseases that trigger irreversible damage to cells in the brain. The most common of these diseases is Alzheimer’s disease (AD), representing 60% to 80% of dementia cases [3]. No treatment has yet been found to prevent or cure cognitive impairment or its progression to dementia and AD [3].

Cognitive impairment, especially difficulty retaining new information, is an early marker of AD. Numerous clinical tests have been validated to evaluate different aspects of cognition that can identify individuals at increased risk of AD within a few years (e.g., rate of learning over time, immediate recall, and delayed recall) [4]. In addition to memory and recall, episodic assessments can be used to assess specific neurocognitive functions: problem‐solving, planning, language, attention, and visuospatial skills. The diagnosis of cognitive impairment and elevated risk of AD is defined by guidelines set by the National Institute on Aging and the Alzheimer’s Association [4]. The guidelines offer flexible assessment criteria to match the assessor’s expertise (general practitioner vs specialist), accessibility to diagnostic resources (e.g., image scanning and blood biomarkers), and the cognitive challenges presented by the patient. In light of this flexibility, there is acknowledgment that the AD population is not homogenous, but a heterogeneous population of subgroups defined by their cognitive abilities and diagnostic criteria [5]. The effect of age on cognitive impairment is undeniable; however, it is important to evaluate the effect of modifiable environmental factors, such as diet, both on the prevention of cognitive decline and onset of AD. Indeed,

identifying cardinal risk factors that commonly contribute to early cognitive impairment and risk of AD, could help (i) identify individuals at high risk of cognitive impairment, (ii) project an individual’s progression toward AD, and (iii) inform strategies to delay the onset of AD.

There is general agreement across studies and study design, of an association between healthy diets, cognitive function, and risk of AD [6–9]. That is, prudent and ‘Mediterranean’ diets, characterised by greater intake of whole grains, fresh fruits and vegetables, fish, and polyunsaturated fatty acids, associate with reduced risk of cognitive impairment, compared to ‘Western’ diets characterised by processed foods, saturated fats, refined grains, and added sugars [10]. PUFAs, lipidic molecules defined by the presence of more than one double bond in the aliphatic chain, are typically more abundant in healthy ‘whole food’ diets, such as those consumed in prudent and ‘Mediterranean’ diets. The two main series of PUFAs are omega-3 (ω3): α-linolenic acid (ALA, C18:3), docosahexaenoic acid (DHA, C22:6) and eicosapentaenoic acid (EPA, C20:5); and omega-6 (ω6): linoleic acid (LA, C18:2) and arachidonic acid (ARA, C20:4) [11].

Precursor ω6 and ω3 PUFAs, LA and ALA, are essential fatty acids determined by diet as they cannot be endogenously synthesised by humans [12], while long-chain (LC) PUFAs (e.g., EPA, DHA, and ARA) may be either endogenously synthesised from their precursor ω3 or ω6 PUFA or consumed directly through diet or supplementation.

ALA, found in high concentrations in chia seeds, walnuts, and oils of flaxseed, canola and soybean [13], can be synthesised by enzymatic desaturation and elongation to supply the body with EPA and DHA [14]. Delta-6 desaturase, the enzyme required in the first rate-limiting stage of ALA conversion [11], is also implicated in the metabolism of LA, meaning that both PUFAs compete for desaturation [13]. The rate of ALA conversion is, therefore, generally low at less than 5% [15]. Food sources of EPA and DHA notably consist of fish and seafood products, including fish oils [16]. Mean intakes of EPA and DHA through food in the UK have been estimated at 244 mg/day [17], substantially lower than the recommended 450 mg/day as stipulated by the Scientific Advisory Committee on Nutrition [18]. However, PUFA intake is perhaps best predicted by the ω6:ω3 ratio present in the diet, rather than the intake of a single PUFA. While ratios between 1:1 and 5:1 have demonstrated positive effects in ameliorating general disease burden [19], the modern Western diet, characterised by a high consumption of meat, vegetable oils, and processed foods, is often reported to have an ω6:ω3 in excess of 15:1 [20]. The chronic low-grade inflammation induced by this

aspect of diet may have consequences for neuronal health, generating an environment that impairs cognitive function and fosters AD pathology [21]. Structurally, ω3 and ω6 LC PUFAs comprise around 30–35% of fatty acids in the brain [11] and have been shown in pre-clinical and human studies to exert effects on metabolic processes involved in brain development and function throughout the life course [11, 22, 23] through modulation of membrane fluidity, gene expression, and inflammation.

while pre-clinical AD models suggest a causal pathway connecting decreased dietary ω6:ω3 ratios with increased PPAR signalling and improved cognitive measures [35, 36]. Collectively, this positions modification of dietary LC PUFAs as a viable approach to improve cognitive function in AD; however, heterogeneity of effect sizes in human trials [37] challenges the generalisability of results.

The fatty acid composition of cell membranes affects the membrane fluidity index by influencing membrane packing [11]. The abundance of LC PUFAs within the phospholipid fractions of the brain, predominantly DHA and ARA [14], increases the fluidity of neuronal membranes [12], which in turn modulates the function of transmembrane and peripheral proteins, such as receptors, enzymes and ion channels involved in vital cellular processes [11]. Following a decrease in absorption across the blood–brain barrier and changes to fatty acid metabolism, the aging brain sees a decrease in membrane LC PUFA concentration [12] and a corresponding decline in membrane fluidity—an observation documented in patients with clinically diagnosed cognitive impairment and AD [24, 25]. Interestingly, aged-related membrane rigidity has been reversed in the hippocampus of pre-clinical AD models through a DHA-enriched diet [26] but its effect on cognitive function in individuals already diagnosed with AD is inconsistent [27, 28].

LC PUFAs are implicated in the expression of several genes [29]. Targets for modulation are primarily from the nuclear receptor superfamily, namely, the retinoid X receptor, retinoic acid receptor and peroxisome proliferator-activated receptor (PPAR) [22]. They function as ligand-activated transcription factors in retinoid signalling [29], which are associated with synaptic plasticity, subsequent memory and learning ability [22] and, therefore, AD disease pathologies [30]. Interestingly, LC ω3 PUFAs show a greater potency in modifying nuclear receptor gene expression compared to LC ω6 PUFAs [31]. PPAR can bind to multiple fatty acids and their derivatives; however, PUFAs, in particular EPA, show superior binding affinity compared to ω6 PUFAs and saturated fatty acids [32]. Furthermore, increased ω3 intake may increase transthyretin [29], a thyroid transport protein that acts as a scavenger of β-amyloid (Aβ) protein fragments [14]. The aggregation and deposition of Aβ42 fragments as neuropathological plaques is considered a primary hallmark of AD [33]. A human model of AD has shown that over-expression of transthyretin supresses Aβ plaque formation, putatively through sequestration [34],

Eicosanoids are a group of inflammatory mediators that include prostaglandins, thromboxanes and leukotrienes [38, 39]. PUFAs act as precursors to the synthesis of eicosanoids through metabolism by cyclooxygenase (COX) and 5-lipoxygenase (5-LOX) enzymes [22]. ARA, a 20-carbon LC ω6 PUFA, is in plentiful supply in neural cell membranes, making it the primary substrate for the majority of eicosanoid production [38]. ARA-derived eicosanoids, which include the

1. 2-series of prostaglandins and thromboxanes, as well as the 4-series of leukotrienes [22], potentiate autocrine and paracrine inflammation [40], with in vitro evidence suggesting they induce Aβ plaque formation [31]. Conversely, EPA, a 20-carbon LC ω3 PUFA, competes with ARA for COX and 5-LOX enzymes [41], thereby lessening ARA metabolism and levels of ARA-derived eicosanoids [38, 42]. In addition, EPA is a precursor for lesser pro-inflammatory eicosanoids, namely, the
2. 3-series of prostaglandins and thromboxanes, and the 5-series of leukotrienes [22, 43], and (along with DHA) anti-inflammatory resolvins and neuroprotectins [38]. Of particular interest is neuroprotectin D1, which can inhibit COX2 expression, reduce ARA metabolism [44], and modulate Bcl-2 expression towards an anti-apoptotic state; thereby, ameliorating the deleterious effects of inflammation [45].

In summary, evidence from molecular, epidemiological, and preclinical and human studies elucidates dietary LC ω3 PUFAs as mediators of cognitive decline and risk of AD; however, a summary of current literature is required to critique and evaluate the strength of existing evidence linking dietary LC ω3 PUFAs with cognitive decline and explain the heterogeneity observed in study results. By understanding the primary mechanisms that contribute to the development and progression of cognitive decline and AD, we are well positioned to offer an evaluation of this evidence. In doing so, we aim to determine the role of dietary and supplemented LC ω3 PUFAs as moderators of cognitive decline and risk of AD.

PubMed and Scopus databases were searched, using “omega-3 fatty acids”, “Alzheimer’s disease” and “cognition” as key search terms, as well as variants of these (“n-3

fatty acids, “PUFAs”, “AD”). Inclusion criteria were: (i) original randomised controlled trials (RCTs) or observational studies; of (ii) human participants ≥ 50 years; (iii) investigating the association between dietary or supplemented ω3 PUFAs; on (iv) qualitative or biological assessment of cognitive function; (v) as a marker of AD risk. Exclusion criteria were: (i) in vitro or (ii) preclinical studies, and (iii) non-original or (iv) non-peer-reviewed publications. No exclusion regarding initial publication date was set and all studies published up to March 2020 were eligible. Identified studies were then compiled into tables to assess and compare the participant sample profile, methods, and outcomes in a clear and concise manner (see Tables 1, 2).

Eight observational studies and 25 RCTs were selected for review (see Supplementary Fig. 1).

The data from observational studies generally support an inverse association between dietary LC ω3 PUFA intake and cognitive decline and risk of AD (Table 1). The Rotterdam Study (1997) is the earliest account in this review of the link being demonstrated in a population-based cohort of elderly subjects (≥ 55 years). Data from this study showed a relative risk (RR) of AD at 0.3 (95% CI: 0.1–0.9) with daily fish consumption above 18.5 g [46]. Interestingly, a follow-up study in 2009 reported contrasting results [47]. Fish, total ω3, EPA and DHA intake of 5395 subjects did not associate with AD [47]. The 10-year follow-up period may have meant that baseline dietary information was not representative of the participants’ diets throughout the duration of the study; however, this does not rule out the potential relevance of earlier dietary habits in AD development.

Further studies have reported similar results [48–51], including a smaller cohort used in the Chicago Memory and Aging Project (n 815), wherein consumption of one or more fish meals per week generated a RR of 0.4 (95% CI: 0.2–0.9), adjusted for age, sex, race, education, energy intake and presence of the ApoE ɛ4 polymorphism [49]. Although this result was near borderline significance (p = 0.07), the inverse, linear association between total ω3 PUFA intake and AD showed clear significance (p = 0.01) [49]. This relationship was also seen for DHA intake specifically, with those in the top three quintiles of intake showing multivariate-adjusted RR as low as 0.2 (95% CI: 0.1–0.8) [49]. As less than 1% of the cohort reported taking ω3 PUFA supplements, it can be deduced that this intake was primarily achieved through diet.

A later prospective study by Schaefer et al.[51] analysed baseline plasma DHA in participants of the Framingham Heart Study and found that those in the upper quartile (mean 3 servings of fish/week) had an all-cause dementia RR of 0.53 (95% CI: 0.29–0.97) compared to those in the lower three quartiles. Although of similar magnitude, the association between plasma DHA and AD was not significant (RR = 0.61; 95% CI: 0.31–1.18; p= 0.14) when upper and lower quartiles were compared [51]. The dietary intake of DHA was also high for these subjects in the cohort, at 1800 mg/day. Fish intake was significantly associated with plasma DHA; however, once adjusted for plasma DHA, the association between dietary fish and DHA intake and AD was not significant [51]. This suggests that DHA is the primary component of fish effecting AD risk and underlines the importance of DHA bioavailability and absorption into the blood, which can be improved by co-ingestion of a high fat meal alongside supplement use [52]. In addition, variability of fish DHA content (% fatty acids) between species and global region [53, 54], is likely to contribute to the heterogeneity of results we observed between observational studies and suggests that subgrouping fish intake by its DHA content may help to reduce this heterogeneity.

A more recent community-based cohort study, the Rush Memory and Aging Project, identified a decreased risk of AD in participants with the highest adherence scores to MIND (Mediterranean-DASH Intervention for Neurodegenerative Delay), DASH (Dietary Approaches to Stop Hypertension) and Mediterranean diets, with hazard ratios (HR) at 0.47 (95% CI: 0.26–0.76), 0.61 (95% CI: 0.38–0.97) and 0.46 (95% CI: 0.26–0.79), respectively [55]. In particular, the MIND diet showed a HR of 0.65 (95% CI: 0.44–0.98) even with moderate adherence [55]. In addition to adequate intake of fish, the diet consists of 15 ‘brain-healthy’ food aspects, including increased fruit and vegetables and reduced red meat and butter, which make it difficult to deduce the role of LC ω3 PUFAs in ameliorating AD risk from these results. These findings agree with previous work investigating the combined effects of multi-nutritional LC ω3 PUFAs-rich drinks [56] on cognitive function and AD risk and encourage future studies to identify novel LC ω3 PUFA-nutrient interactions or nutritional patterns that might offer even greater protection against cognitive impairment and future AD risk.

The same cohort was tested for cognitive ability using a 21-test battery to assess global cognition and five specified cognitive domains, including episodic memory and visuospatial ability [57]. Although ω3 PUFAs from food sources were not associated with cognition, ω3 supplementation was significantly associated with slower rates of decline in global cognition and episodic memory scores compared to non-consumers [57]. This suggests that an increased habitual intake of ω3 PUFAs reliably delivers an adequate dosage to

Table 1 Observational studies investigating the association between dietary ω3 intake and AD risk

References Cohort Year Study characteristics Follow-up Dietary recall Findings

1. [46] Rotterdam Study 1997 n 5386 Age: ≥ 55 years
2. [47] 2009 n 5395 Age: ≥ 55 years
3. [48] Cardiovascular Health Cognition Study
4. [49] Chicago Memory and Aging Project
5. [50] Hordland Health Study 2007 n 2031 Age: 70–74 years
6. [51] Framingham Heart Study 2006 n 899 Age: 55–88 years

[55] Rush Memory and Aging Project

2015 n 923 Age: 58–98 years

[57] 2016 n 915 Mean age: 81.4 years

9.1 years Semi-quantitative FFQ Significant 47% reduction in risk of developing all-cause dementia in upper quartile of plasma DHA content (mean DHA intake = 0.18 g/d, mean fish intake = 3 servings/week) (95% CI: 0.29–0.97). No significant reduction in risk of AD specifically

4.5 years Semi-quantitative FFQ Moderate and high adherence to MIND diet showed lower risk of AD compared to first tertile. High adherence to DASH and Mediterranean diet also associated with lower AD risk

4.9 years Intake of food sources of ω3 PUFAs not associated with cognitive decline, however, fish oil supplement consumers had slower rates of decline in global cognition and episodic memory measures than non-consumers

FFQ food frequency questionnaire

supplementation Exposure periodOutcome MeasuresFindings

ReferencesStudy designYearSampleDiagnostic approachDose and method of

Clinical trials investigating the association between supplementary ω3 intake and cognitive functionTable 2

non-significant increase

but not observed in AD

ADAS-cog parameters.

Plasma EPA associated

score in ω3 MCI group

nition between groups.

not cognitive function.

cant difference in cog-

mild cognitive decline

cognitive decline rate.

5.5 monthsCIBIC, ADAS-cogImprovement in general

No significant effect on

No significant effect on

DHA and decrease in

compared to placebo,

in cerebrospinal fluid

12 monthsMMSE, ADAS-cogNo statistically signifi-

plasma transthyretin,

No associations with

rate based on several

Significant improve-

Significant increase in

showed a significant

clinical function but

association between

with slower decline

ment in ADAS-cog

A subset with very

changes in plasma

decrease in rate of

Mean age: 74 Significant positive

cognitive decline

transthyretin

level of AD

biomarkers

biomarkers

group

TNF-α, sIL-1RII) and

F2-isoprostane, 8-iso-

12 monthsTransthyretin in plasma

PGF2α and 15-keto-

(tau, hyperphospho-

fluid as indicated by

ers in plasma (IL-6,

cerebrospinal fluid

6 monthsInflammatory mark-

and cerebrospinal

rylated tau, Aβ)42

Mean age: 70.5 Change in levels of

dihydro-PGF2α

nephelometry

1700 mg/day DHA and

1080 mg/day EPA as

720 mg/day DHA and

600 mg/day EPA as

Placebo: olive oil

Placebo: corn oil

capsules

capsules

capsules

capsules

history, psychometric

DSM-IV interview and

testing, blood analy-

medical assessment

DSM-IV, medical

by psychiatrist or

Clinical diagnosis:

neurologist

ses, MRI

[81]2014Moderate AD patients

[84]2013Mild to moderate AD

trolled trial 2008Mild to moderate AD

[80]2009Mild to moderate AD

2006Mild to moderate AD

[59]2015AD patientsn174

Mean age: 75.1

Mean age: 75

Mean age: 70

Mean age: 74

patients

patients

patients

or MCI

n174

n174

n40

n29

n35

[60]Randomised double-

[58]Randomised double-

blind placebo-con-

controlled trial

blind placebo-

(OmegAD)

and was associated with

verbal memory, delayed

cantly fewer PAL errors

ing memory, immediate

plasma DHA levels that

in short-term and work-

change in memory over

in scores for Full Scale

Mean age: 73.6 24 monthsSignificant improvement

Significant improvement

correlated with signifi-

for intervention group

Information and Digit

measures of cognitive

Intelligence Quotient,

Intelligence Quotient,

Recognition Memory

WAIS-R, MRI Increased hippocampal

Span for intervention

RBANS test Significantly improved

compared to placebo

improved immediate

cant improvement in

scores for Full Scale

recall capability and

lesions compared to

in MCI and organic

immediate memory

tion and Digit Span

volume and signifi-

and delayed Verbal

Verbal Intelligence

placebo, not in AD

group compared to

Quotient, Informa-

Two-fold increase in

brain atrophy No effect on any

12 months

placebo

decline

supplementation Exposure periodOutcome MeasuresFindings

of Cambridge, Spatial

Recognition Memory,

ciate learning, Verbal

3 monthsJapanese version of the

5.5 monthsCANTAB paired asso-

battery comprised of

12 monthsChinese version of the

Pattern Recognition

Memory, Stockings

boxes CDR, rate of

WAIS-R, CDT and

18 monthsADAS-cog, sum-of-

Working Memory,

WMS-R, RAVLT,

12 monthsNeuropsychological

components from:

WAIS-III

MMSE

Placebo: corn or soy oil

derived from algae as

derived from algae as

and 450 mg/day EPA

240 mg/day DHA as

Placebo: corn and soy

assessed by authors 240 mg/day ARA and

derived from fish as

Placebo: soybean oil

900 mg/day DHA as

ReferencesStudy designYearSampleDiagnostic approachDose and method of

2000 mg/day DHA

2000 mg/day DHA

psychologists 1300 mg/day DHA

Placebo: olive oil

Placebo: corn oil

capsules

capsules

capsules

capsules

capsules

capsules

capsules

capsules

capsule

oil

assessment by clinical

MCI, medical history,

criteria for AD inci-

Mean age: 76 Alzheimer’s Disease

NINCDS-ADRDA

Cooperative Study

Mean age: 64.9 Neuropsychological

clinic institutional

Petersen’s criteria

Mean age: 74.5 Clinical diagnosis

Petersen’s criteria

by neurologist:

review boards

dence

trolled trial 2010Age-related cognitive

trolled trial 2010Mild to moderate AD

controlled trial 2006MCI, organic brain

Mean age: 64.9

lesions, AD

Mean age: 70

decline

n219

n217

n295

n485

trolled trial 2017MCI

[61]2018MCI

blind placebo-con2012MCI

n35

n39

[64]Randomised placebo-

[28]Randomised double-

[27]Randomised double-

[62]Randomised double-

[63]Randomised double-

blind placebo-con-

blind placebo-con-

blind placebo-con-

other tests, although all

tion with multi-domain

placebo. No significant

difference in scores for

observed in any cogni-

tive function measures

group showed reduced

No significant effect on

[67]2018No significant effect on

40 monthsMMSENo significant effect on

No significant effect on

composite score with

scores lower in inter-

or without combina-

No significant effects

scores compared to

decline in COWAT

ω3 supplementation

composite scores

battery test score

vention group

MMSE score

intervention

supplementation Exposure periodOutcome MeasuresFindings

bol Substitution Test,

COWAT, Digit Sym-

delayed and recogni-

MMSE World Back-

4 monthsMMSE Serial Sevens,

tion verbal memory

Reminding Test, 10

and Cued Selective

fluency, alternating

for: TICS-M, letter

fluency, WMS-III,

total recall of Free

wards, immediate,

ing, delayed recall

including free and

Category Naming

Backward Count-

items on MMSE,

fluency, category

Test, CDR, TMT

60 monthsComposite scores

of TICS-M and

36 monthsBattery of tests

WMS-III

AREDS formulation

625 mg/day DHA and

MMSE 800 mg/day DHA and

N/A160 mg/day DHA and

N/A350 mg/day DHA and

or without 2000 mg

600 mg/day EPA as

225 mg/day EPA as

240 mg/day EPA as

650 mg/day EPA as

ReferencesStudy designYearSampleDiagnostic approachDose and method of

Placebo: paraffin oil

with fish oil, with

margarine treated

Placebo: standard

Placebo: standard

Placebo: olive oil

margarine

capsules

capsules

capsules

capsules

capsules

ALA

Mean age: 71.1 Medical and neuropsy-

MRI, blood analyses

from memory clinic

general practitioner,

Mean age: 75.3 Medical history from

referral, NINCDS-

chological history

ADRDA criteria,

ment, adult macular

[71]2017Lowest quartile of ω3

Omega Trial) 2012No cognitive impair-

blind placebo-con2015No cognitive impair-

trial (MAPT) 2017Memory complaints

Memory complaints

trolled trial 2015CIND, early AD

Mean age: 69.1

Mean age: 72.7

degeneration

Mean age: 76

index

n2911

n1680

n3424

ment

n183

n76

controlled trial (Alpha

controlled superiority

[68]Double-blind placebo-

[66]Randomised placebo-

[65]Randomised double-

[69]Randomised double-

blind placebo-con-

Increase in cerebral blood

in omega-3 intervention

MMSE or AMT scores

flow and volume larger

No significant difference

and significantly more

OLM scores observed

6 monthsMMSE, AMTNo significant effect on

WMS-R compared to

more significant with

in isoprostane levels.

Memory Significant increase in

Effects on cognition

on composite score,

addition of LA than

cog, CIBIC Fewer reduced scores

decline much lower

than placebo group

in supplementation

improved scores in

group compared to

although cognitive

Test Battery No significant effect

omega-3 s alone

than expected

placebo

placebo

supplementation Exposure periodOutcome MeasuresFindings

3 monthsWMS-R delayed verbal

MMSE, ADL/IADL,

indicated by cerebral

Neuropsychological

6 monthsCerebral perfusion as

blood flow and cer-

ebral blood volume

recall task, ADAS-

24 monthsModified version of

12 monthsChange in levels of

F2-isoprostane,

6 monthsObject Location

ADAS-cog

day EPA derived from

ω3 placebo: soybean oil

nutrition combination

nutrition combination

1200 mg/day DHA and

teria, MMSE, MRI 1200 mg/day DHA and

1320 mg/day EPA as

fish plus 600 mg/day

capsules; or 675 mg/

Placebo: sunflower oil

Placebo: sunflower oil

teria, MMSE, CDR 675 mg/day DHA and

specialist 880 mg/day DHA and

N/A880 mg/day DHA and

180 mg/day DHA and

cod liver oil capsule

derived from fish as

derived from fish as

300 mg/day EPA as

120 mg/day EPA as

300 mg/day EPA as

alpha lipoic acid as

as capsule with 5%

in Souvenaid drink

in Souvenaid drink

Placebo: coconut oil

day DPA, 975 mg/

ReferencesStudy designYearSampleDiagnostic approachDose and method of

1320 mg/day EPA

Fortasyn Connect

Fortasyn Connect

Placebo: isocaloric

Placebo: isocaloric

975 mg/day EPA

capsules

capsules

capsules

capsules

capsule

fish oil

tablet

drink

drink

Mean age: 66.5 Mayo clinic criteria for

Mean age: 75.9 NINCDS-ADRDA cri-

Mean age: 73.7 NINCDS-ADRDA cri-

cose PET analysis by

to intervention using

18F fluorodeoxyglu-

criteria, CSF, MRI,

and screening prior

MCI by neurology

12-month follow-up

DSM-IV, MMSE

Mean age: 71 NINCDS-ADRDA

clinician

trolled trial 2014Mild to moderate cog-

controlled trial 2016No cognitive impair-

nitive impairment

Mean age: 74.6

trolled trial 2017Prodromal AD

controlled trial 2014Probable AD

Age: 50–70

trolled trial 2010Mild AD

ment

n199

n225

n311

trolled trial 2017MCI

n39

n13

n44

[70]Double-blind placebo-

[78]Randomised placebo-

[82]Randomised double-

[72]Randomised double-

[73]Randomised double-

[74]Randomised double-

blind placebo-con-

blind placebo-con-

blind placebo-con-

blind placebo-con-

increased significantly

in MCI but not in AD.

No other significant

Aβ phagocytosis

associations

supplementation Exposure periodOutcome MeasuresFindings

matory genes (as indi-

tion (as indicated by

resolvin-D1 produc-

enzyme immunoas-

scription of inflam-

cated by RC-PCR),

17 monthsAβ phagocytosis (as

microscopy), tran-

indicated by flow

cytometry and

say), MMSE

10ug vitamin D3 and

drink (also contained

ReferencesStudy designYearSampleDiagnostic approachDose and method of

EPA as Smartfish

1000 mg/day DHA

and 1000 mg/day

resveratrol)

criteria, Petersen’s

Mean age: 77.4 NINCDS-ADRDA

criteria for MCI,

MMSE

a2015MCI, AD[83]Open study

n29

(continued)Table 2

DSM-IVdiagnostic and statistical manual of mental disorders fourth edition,NINCDS-ADRDANational Institute of neurological and communicative disorders and stroke and the Alzheimer’s

aNot classified as a clinical trial but was included in the table for the purpose of comparison to other biomarker studies

disease and related disorders association

influence cognitive outcomes. Several of these studies also identified a significant relationship between increased ALA consumption and reduced AD risk, but interestingly only in ApoE ɛ4 carriers [49, 57].

In summary, observational studies support the notion that LC ω3 PUFAs offer a protective effect against cognitive decline and risk of AD. The period of study in observational studies is far longer than can be realistically achieved in a clinical setting, meaning that prolonged exposure to dietary LC ω3 PUFAs can be represented in analysis, in comparison to a transient response to increased LC ω3 PUFA intake. However, the observational studies identified in this review were conducted on European and American populations, meaning that these data may not be generalisable to other populations with varying ratios of ApoE polymorphism, and differing quantities and sources of fish and LC ω3 PUFA consumption.

Future epidemiological studies are required to evaluate the distinct and interactive associations between LC ω3 PUFAs and foods they are commonly consumed alongside (i.e., as in a Mediterranean diet) and the effect of cooking methods on the efficacy of LC ω3 PUFAs. Concurrent dietary parameters such as increased fruit and vegetable intake and decreased saturated fat intake may also play a role in cerebral health and support the metabolic effects of LC ω3 PUFAs.

The OmegAD trial [58] was the first large, randomised placebo-controlled trial to investigate the effects of ω3 PUFAs on the cognitive ability of AD patients (Table 2). Subjects were randomised to receive either ω3 PUFA capsules (1700 mg/day DHA and 600 mg/day) for the full 12 months of the study, or 6 months of placebo capsules followed by 6 months of ω3 PUFA capsules. Scores for the Mini-Mental State Examination (MMSE) and the cognitive portion of the Alzheimer’s Disease Assessment Scale (ADAS-cog) were taken at baseline, 6 and 12 months; however, no statistically significant differences in scores between the two groups were established [58]. A subgroup of 32 patients with very mild AD (MMSE > 27 points) showed a significant attenuation of cognitive decline compared to placebo, as indicated by MMSE score after 6 months of intervention [58].

A further study as part of the OmegAD trial found that the diminution of cognitive decline as measured by ADAScog scores was significantly associated with increasing plasma DHA, and that plasma EPA was associated with a number of parameters within the ADAS-cog [59]. Similar conclusions can be drawn from a shorter trial by Chiu et al. [60], assessing a formulation with lower DHA and higher

EPA content than the OmegAD trial (DHA:EPA 2:3 [60] vs 3:1 [58]). Although the mixed model indicated a positive change in global clinical function in the intervention group, as measured by Clinician's Interview-Based Impression of Change (CIBIC), the improvement of ADAS-cog score over the intervention was selective to those with mild cognitive impairment (MCI); no effect was seen in AD patients. Intriguingly, this study reported lower levels of plasma ARA in the ω3-treated group compared to the control [60], which may suggest a decrease in pro-inflammatory eicosanoid synthesis caused by the inhibitory action of EPA on ARA metabolism. Increased EPA on erythrocyte membranes was associated with improved ADAS-cog scores [60].

Zhang et al. [28] investigated the effect of high dose algal-derived DHA on cognitive function of those with MCI, defined as subjective memory complaints accompanying a score of 1.5 standard deviations below age- and education-matched controls in MMSE memory subtest, but without the presence of AD or related diseases [28]. Following 12 months of supplementation with 2000 mg DHA/ day, the intervention group achieved significantly higher test scores for Full Scale Intelligence Quotient and subdomains of Information and Digit Span [28]. In a follow-up paper after 24 months of supplementation, the authors reported the same significant improvements as seen after 12 months, with an additional improvement in Verbal Intelligence Quotient [61]. This dosage has also been trialled in patients with mild to moderate AD (MMSE 14–26), but there was no significant effect on rate of cognitive decline as indicated by ADAS-cog and Clinical Dementia Rating (CDR) [27]. It may be important to note that one of the placebos used in these studies was soybean oil, known to contain, amongst other fatty acids, a significant amount of ALA [13], which may have compromised the estimate of the effect size. Indeed, as a precursor for LC ω3 PUFA, it is likely that a small percentage of ALA was synthesised into EPA or DHA and contaminated the control group; however, without another control group, the magnitude of contamination cannot be determined.

The efficacy of LC ω3 PUFA supplementation on MCI has been further investigated by a number of trials [62 65]. Yurko-Mauro et al. [62] examined DHA supplementation in subjects with age-related cognitive decline (MMSE > 26). After 24 weeks of treatment (900 mg/day), the intervention group had significantly fewer Paired Associate Learning pattern errors than placebo subjects, as well as improved immediate and delayed Verbal Recognition Memory [62]. These changes were significantly associated with increased plasma DHA levels, of which the intervention group exhibited a twofold rise in plasma DHA [62]. Improved immediate verbal memory has also been reported in MCI patients treated with DHA-EPA combination supplements [63] and DHA supplements at a substantially lower dosage [64]. Interestingly, this study [64] delivered 240 mg/day DHA alongside

an equivalent dose of ARA. Increased concentration of ARA in neuronal membranes may have increased the fluidity index of brain cells in the treated group, consequently leading to improved synaptic function. By contrast, AD subjects did not exhibit any improvement in immediate memory [64]. Similar findings were documented in another trial [65]; however, this study also found no association between LC ω3 PUFA supplementation (625 mg/day DHA 600 mg/day EPA) and cognitive function in cognitively impaired individuals without dementia (CIND), contradicting evidence from other research [62 64]. These studies raise questions regarding the effects of study parameters, such as study duration and dosage, that may have contributed to the results of trials successfully associating LC ω3 PUFAs with improved cognition in MCI. Specifically, they suggest that durations of ≥ 5 months and DHA supplements exceeding 900 mg/day, are required to elicit a significant effect on cognitive ability in elderly populations.

Several large-scale trials (n ~ 1600–3400), including the Multidomain Alzheimer Preventive Trial (MAPT) and Alpha Omega Trial, have found no association between LC ω3 PUFAs and cognition, both in those with memory complaints [66, 67] and those deemed cognitively healthy [68, 69]. Although these studies use large sample sizes and substantial intervention periods, the dosage of LC ω3 PUFAs are comparatively lower than in other trials conducted in this area (Table 2). While these levels better mimic the level of ω3 PUFAs achievable through the diet, the results of studies using higher supplementary doses have more consistently reported significant effects of LC ω3 PUFAs on cognitive function in MCI [28, 61, 63] and at-risk older individuals [70]. Interestingly, a secondary analysis of MAPT study subjects with memory complaints in the lowest quartile of ω3 index revealed an increase in Controlled Oral Word Association Test (COWAT) scores for the intervention group over the 36-month intervention, where the control group experienced a decrease in mean score [71]. Although this was the only finding to achieve statistical significance, the intervention group performed better than the placebo group in all other tests conducted [71]. Interestingly, a smaller 6-month trial (n 199) supports these findings in cases of MCI, reporting no significant change in MMSE or Abbreviated Mental Test (AMT) following ω3 PUFA supplementation [72].

Scheltens et al. [73] randomised patients with mild AD (MMSE 20–26) to receive either 1200 mg/day DHA and 300 mg/day EPA as a medical supplement drink (Souvenaid), or an isocaloric placebo drink, for 12 weeks. A significant improvement was reported in revised Wechsler Memory Scale (WMS-R) delayed recall score, although ADAS-cog and CIBIC scores did not change. A subgroup analysis of patients with very mild AD also achieved significantly improved immediate verbal recall after 12 weeks [73]. A 24-month trial on a similar population with prodromal AD

(mean age 71, mean MMSE 26.7) has since been conducted to establish the longer term influence of Souvenaid [74]. The authors reported a moderately positive effect based on lower rates of cognitive decline in the intervention group than were expected based on projected 24-month decline observed in AD patients [75]. However, no significant effect on the composite score of the neuropsychological test battery used was reported [74]. Souvenaid contains a combination of other nutrients including vitamin C (80 mg) and a number of B-vitamins (3 μg B12, 1 mg B6, 400 μg folic acid). Interactions between B-vitamins and ω3 PUFAs in affecting reduced rates of brain atrophy have been reported in MCI [76]. Although the efficacy of B-vitamin treatment in AD is still yet to be thoroughly established [77], the potential for the effects observed in the Souvenaid trials to be influenced by the additional nutrients in the drink make it difficult to isolate the impact of ω3 PUFAs alone.

The Souvenaid trials raised questions about the combination of ω3 PUFAs with other agents in supplementation. Shinto et al.[78] analysed the effects of ω3 PUFA treatment (675 mg/day DHA 975 mg/day EPA) with and without the addition of α-lipoic acid (600 mg/day), an organosulphur compound and antioxidant implicated in the protection of mitochondria from oxidative damage [79] and reduction of inflammatory markers [78], as is seen in AD. Although no significant difference was found in ADAS-cog scores, Instrumental Activities of Daily Living (IADL) scores of probable AD subjects showed significantly less decline over the 12-month trial period for both ω3 and ω3-α-lipoic acid groups compared to placebo; however, a significant difference between MMSE scores for treatment and placebo groups was only observed in the case of combined ω3-αlipoic acid supplementation [78]. As levels of endogenous α-lipoic acid decrease with age [79], these results suggest the use of lipoic acid in conjunction with ω3 PUFA supplementation may have biological plausibility for cases of late-onset AD, although further research is warranted. In the placebo arm, the authors used a capsule containing soybean oil and 5% fish oil [78]. While the ω3 concentrations may not have been as high as in the intervention arm, the potential influence on results is worth noting.

As well as cognitive testing, numerous studies also measured biological changes as indicators of cognitive function [78, 80 84]. As part of the OmegAD study, 35 AD subjects were randomised to 6-month supplementation of ω3 PUFAs (1700 mg/day DHA 600 mg/day EPA) or placebo [80]. Plasma samples were tested at baseline and 6 months for the inflammatory markers IL-6, tumour necrosis factoralpha (TNF-α) and soluble interleukin-1 receptor type-II (sIL-1RII). Cerebrospinal fluid was extracted by lumbar

puncture at baseline and 6 months and tested for IL-6, TNFα, sIL-1RII, as well as tau protein, hyperphosphorylated tau protein and Aβ peptides, all hallmarks of AD pathology [33]. A significant association between sIL-1RII and Aβ42 was made at baseline; however, no associations related to ω3 PUFAs were found with any of the biomarkers measured [80]. The same sample was examined for changes in urinary F2-isoprostanes and prostaglandin F2α, both formed through COX pathways associated with oxidative stress [81]. 15-keto-dihydro-PGF2α, a metabolite of prostaglandin F2α, was used as a biomarker of inflammation. No changes were observed in any of these biomarkers. Further research on ω3 PUFA supplementation and F2-isoprostane levels in cases of probable AD reached a similar conclusion [78]. However, neither study reported a qualitative assessment of cognitive function at baseline and follow-up, so it is not possible to confirm the null hypothesis or validate the association between biomarkers and cognitive function.

A number of biomarker studies have reported associations between ω3 PUFAs and quantitative indicators of cognitive function [82, 83]. A cerebral perfusion study revealed increased cerebral blood flow and volume in posterior corticoid regions of MCI subjects following high dose ω3 PUFA supplementation (880 mg/day DHA and 1320 mg/day EPA) for 6 months [82]. Cerebral hypoperfusion has been associated with AD as a consequence of neuronal tissue damage and brain atrophy [85]. Although no statistically significant effect of 18-month high dose DHA supplementation was reported in slowing the rate of brain atrophy in cases of mild to moderate AD [62], a more recent study [61] of similar design with a larger sample size (n 219 vs 102 [62]) identified an association between the same DHA dosage (2000 mg/day) and increased hippocampal volume in MCI subjects after 12 months [61]. In another study, administration of a drink containing 1000 mg DHA and 1000 mg EPA amongst other nutrients, including vitamin D and resveratrol (Smartfish), was significantly associated with increased Aβ phagocytosis by monocytes as indicated by flow cytometry in MCI and pre-MCI subjects, but not in AD [83]. These studies further support the efficacy of LC ω3 PUFAs as early interventions and suggest that physical brain measures, while time-consuming and costly, provide a more reliable measure of cognitive degeneration compared to blood and cerebrospinal fluid analytes.

The largest biomarker study identified in this review, with more than 100 subjects, was undertaken by Faxén-Irving and colleagues, as a further part of the OmegAD trial [84]. Transthyretin, a Aβ binding protein thought to reduce plaque formation [86], was the compound of interest. 174 patients with mild to moderate AD completed two consecutive 6-month supplementation periods as two groups—ω3/ω3 and placebo/ω3. Plasma transthyretin levels after 6 months significantly decreased in the placebo group, at which point

a correlation was identified between DHA and transthyretin (rho = 0.17, p = 0.03) [84]. At the 12-month follow-up, after both groups had been supplemented with ω3 PUFAs, plasma transthyretin increased significantly in both groups, at which point a correlation was identified between MMSE scores and transthyretin (rho = 0.16, p = 0.03), alongside an inverse correlation with ADAS-cog scores (rho = − 0.2, p = 0.02) [84]. Cerebrospinal fluid was analysed over 6 months in a subset of 35 subjects; however, the increase in cerebrospinal fluid transthyretin observed in the ω3/ω3 treatment group was not significant [84]. This trial is one of few that offers evidence that ω3 PUFAs may improve cognition in those with established AD.

In controlled trials, participants have usually already received a diagnosis of AD, meaning that ω3 PUFAs are tested for their treatment effect on a disease that has already been established with notably manifested symptoms of memory impairment. Indeed, based on the data reviewed here, wherein the majority of clinical trials that made positive associations between supplementation and cognition were conducted in cases of MCI as opposed to AD, a protective effect of ω3 PUFAs is inferred.

Overall, LC ω3 PUFA supplementation appears to be well-tolerated, even in high doses (Table 2). Although LC ω3 PUFAs may not be a validated treatment strategy for cognitive impairment or AD based on the current evidence, their successful implementation as an early intervention strategy for neuroprotection in MCI and healthy older populations is encouraging. Large-scale trials of long duration (≥ 5 months) using higher dosage supplements (900 mg DHA/day) across diverse populations are required to validate these findings, to improve generalisability and determine safe and optimum dosage. Furthermore, the combination of LC ω3 PUFAs with other nutrients of interest, such as α-lipoic acid and B-vitamins, is an intriguing area of research that may identify additional supplementation strategies and nutrient interactions to prevent and ameliorate cognitive decline in the elderly.

Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s00394-021-02655-4.

Funding MAZ is supported by Wellcome Trust (217446/Z/19/Z).

Cognitive impairment and AD present major challenges for clinical medicine, with multi-faceted pathologies that are still not entirely understood. The role of environmental factors to prevent cognitive impairment and subsequent AD continues to be an area of extensive research, particularly pertinent to late-onset cases. Due to their widely recognised influence on cognitive function, ω3 PUFAs present a compelling opportunity for nutritional therapy in cognitive decline and AD.

Evidence from observational studies appears promising for prevention of cognitive decline. Regular consumption of LC ω3-PUFA rich foods in healthy populations without preexisting AD or dementia suggests a protective effect against future AD. Conversely, clinical trials that primarily focus on participants already diagnosed with AD consistently report no effect. The incongruence between these findings highlights the precedence of LC ω3 PUFA intake earlier in life, before cognitive decline is apparent and AD risk has been recognised.

Although many observational studies adjusted for known confounders, including the ApoE ɛ4 allele, this was not consistent across the literature. Those that specifically analysed ApoE ɛ4 suggest that the metabolic effects of different LC ω3 PUFAs differ between ɛ4 carriers and non-carriers; therefore, more targeted studies within this high-risk population are warranted to better understand ɛ4-specific effects of different types of LC ω3 PUFAs.

Conflict of interest On behalf of all authors, the corresponding author declares no conflict of interest.

Data availability Not applicable. Code availability Not applicable.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.