Dietary pattern, food, and nutritional supplement effects on cognitive outcomes in mild cognitive impairment: a systematic review of previous reviews

Abstract

Context

Nutritional interventions may benefit cognition in people with mild cognitive impairment (MCI). However, evidence is yet to be synthesized in a way that can inform recommendations for clinical and public health settings.

Objective

To systematically review evidence on the effect of dietary patterns, foods, and nutritional supplements on cognitive decline in individuals with MCI.

Data Sources

Guided by the Preferred Reporting items for Systematic Review and Meta-Analysis Protocols 2015 statement, the Medline, EMBASE, and CINAHL databases, the JBI Database of Systematic Reviews and Implementation Reports, Cochrane Database of Systematic Reviews, and Database of Abstracts of Reviews of Effects were searched (publication years 2005 to 2020). Included studies were English-language systematic reviews and meta-analyses of randomized controlled trials and cohort studies reporting on the effectiveness of nutritional interventions on cognition of individuals with MCI.

Data Extraction

Two reviewers independently selected studies and extracted data on cognitive outcomes and adverse events. Review quality was assessed using AMSTAR 2 (A Measurement Tool to Assess Systematic Reviews–2). Primary study overlap was managed following Cochrane Handbook guidelines.

Data Analysis

Of the 6677 records retrieved, 20 reviews were included, which, in turn, reported on 43 randomized controlled trials and 1 cohort study that, together, addressed 18 nutritional interventions. Most reviews were limited by quality and the small number of primary studies with small sample sizes. Reviews were mostly positive for B vitamins, omega-3 fatty acids, and probiotics (including 12, 11 and 4 primary studies, respectively). Souvenaid and the Mediterranean diet reduced cognitive decline or Alzheimer’s disease progression in single trials with <500 participants. Findings from studies with a small number of participants suggest vitamin D, a low-carbohydrate diet, medium-chain triglycerides, blueberries, grape juice, cocoa flavanols, and Brazil nuts may improve individual cognitive subdomains, but more studies are needed.

Conclusions

Few nutritional interventions were found to convincingly improve cognition of individuals with MCI. More high-quality research in MCI populations is required to determine if nutritional treatments improve cognition and/or reduce progression to dementia.

Systematic review registration

Open Science Framework protocol identifier DOI:10.17605/OSF.IO/BEP2S.

INTRODUCTION

Dementia and cognitive decline are a growing global health challenge, primarily due to the increased longevity of the world’s population.^1,2 Cognition declines naturally with age; however, when more significant memory loss occurs than with normal aging, mild cognitive impairment (MCI) may be diagnosed.³ MCI is of particular concern because 46% of individuals with MCI develop dementia within 3 years.⁴ With >152 million adults worldwide projected to develop dementia by 2050, there is an urgent need to mitigate the risks for progression of this disease and manage the significant economic and social burdens on patients, families, and healthcare systems.^5,6

There is currently no cure for dementia, and strategies to reduce the burden of disease are focused on managing modifiable risk factors and slowing the progression of cognitive decline.¹ Despite the recent US Food and Drug Administration approval of aducanumab as a therapy for Alzheimer’s disease (AD),⁷ increasing evidence has emerged about the protective role of nutritional interventions to promote healthy brain aging and prevent cognitive decline in older adults.^8,9 Therefore, targeting dietary approaches provides a viable, cost-effective way to prevent or slow neurodegeneration.

Whole-diet approaches have been advocated to delay or prevent cognitive decline, with the Mediterranean diet being the most extensively studied. The Scientific Advisory Committee on Nutrition undertook a review in 2018 and although the group found limitations in the evidence base, it also found support for the Mediterranean diet.⁹ The World Health Organization (WHO) recommends greater adherence may be associated with decreased risk of MCI and AD,¹ most likely through its composition of high-antioxidant–containing and anti-inflammatory foods.^10,11 However, more research is needed to determine effects within MCI populations, because findings remain discordant between Mediterranean and Western dietary environments.¹² The benefits ascribed to the Mediterranean diet parallel other dietary approaches, such as the Dietary Approaches to Stop Hypertension (DASH) diet,^13–15 which also show promising effects for improving cognition.^10,11 These 2 diets have been combined to create the brain-health specific Mediterranean–DASH Diet Intervention for Neurodegenerative Delay (MIND), which has positive effects on cognition but has not yet been reviewed within MCI populations.^1,16 There is also evidence to suggest the efficacy of the ketogenic diet for individuals with cognitive decline, most likely by compensating for reduced glucose uptake in the brain evident in MCI.^17,18 The Modified Mediterranean–Ketogenic Diet is a new combined diet therapy that may improve AD biomarkers by altering the gut microbiome and metabolites.¹⁹ Reviews have examined the efficacy of these dietary approaches in healthy populations and individuals with dementia; however, findings for their association with cognitive function in MCI populations is limited.²⁰

Although evidence suggests that modifying several aspects of dietary intake simultaneously is more likely to promote better cognition than supplementation with only some nutrients,¹ individual foods and nutrients have also been implicated in cognitive health outcomes. The antioxidant vitamins C and E, the B vitamins, and omega-3 polyunsaturated fatty acids (PUFAs) have some supporting evidence for reducing cognitive decline.^21–23 Vitamin A and carotenoids have shown positive effects on AD biomarkers, inhibiting the formation of β-amyloid proteins; however, few studies exist in individuals with MCI.²⁴ Antioxidant-rich foods, including fruits, vegetables, and nuts, may play a role in decreasing dementia risk by reducing oxidative stress in the brain,^25,26 and greater fish consumption has reduced memory decline in healthy participants, but not those with MCI.²⁷ Overall, there is little evidence to confirm whether these nutrients and foods alone will reduce the progression of MCI to dementia.

To our knowledge, no overview of reviews has been conducted to summarize the large amount of evidence on the effect of nutritional interventions on individuals with MCI. The aim of this systematic review of systematic reviews was to examine the existing evidence from relevant reviews, rather than primary studies, on the effect of dietary patterns, foods, and nutritional supplements to support cognitive function for individuals with MCI.

METHODS

Protocol and registration

This overview was conducted according to Open Science Framework protocol identifier DOI:10.17605/OSF.IO/BEP2S and was guided by the Preferred Reporting items for Systematic Review and Meta-Analysis Protocols (PRISMA) 2015 statement (Tables S1 and S2, Supporting Information online).²⁸ Conduct and reporting were developed according to the Joanna Briggs Institute methodology for systematic reviews.²⁹ Piloting was conducted by reviewers at each stage of the study selection, data extraction, and quality assessment.

Eligibility criteria

Inclusion criteria

Included studies were systematic reviews that met the eligibility criteria specified in Table 1.

Table 1.

Inclusion criteria for studies in the current review

Criteria	Description
Study design	Systematic reviews or meta-analyses of randomized control trials or cohort studies for which a quality assessment was performed by review authors
Population	Human adults with mild cognitive impairment (diagnosed through any recognized diagnostic criteria).
	A review was only included if data on individuals with mild cognitive impairment could be separated from the total population.
Settings	All settings
Interventions/exposure	Dietary patterns, foods, dietary nutrients, vitamins, and minerals in the form of supplements that are normally available through the diet Exposure was nutritional intake categorized on the basis of dietary intake assessment. Reviews were only included when the nutritional intervention could be separated from other interventions.
Comparisons	Placebo/sham, alternative or no treatment, usual practice, nonexposed group
Outcomes	Primary: cognitive function (measured through recognized cognitive tests). Results must be extractable, either in the results summary table or within text. Secondary: Adverse effects
Length of follow-up	No limits were placed on the follow-up period.

Exclusion criteria

Reviews were excluded if they included populations that had other conditions associated with cognitive impairment (eg, subjective MCI, dementia, brain injury associated cognitive impairment, other neurological conditions). Studies published in languages other than English were excluded.

Search methods

A systematic literature search was conducted in the Medline, Embase, and CINAHL databases, the JBI Database of Systematic Reviews and Implementation Reports, the Cochrane Database of Systematic Reviews, and Database of Abstracts of Reviews of Effects up to September 7, 2021. Only articles published after 1990 were included, because few reviews were published before this time.³⁰ The search was limited to published systematic reviews and meta-analyses. The search strategy was developed using Medline and adapted to other databases. The search strategy combined broad keywords and Medical Subject Heading terms for the intervention (ie, nutrition, diet, eating, nutritional supplements, dietary pattern, nutritional pattern), outcomes (cognitive impairment, cognitive decline, cognitive disorder, memory loss, cognitive ability loss, dementia, AD), and type of study (systematic review, meta-analysis). No search limits were placed on population or comparators, to increase the scope of the results. The full search strategy is available in Table S5 in the Supporting Information online.

Study selection

The titles and abstracts of search results were independently screened by 2 authors (G.Z. and V.A.) for potential studies for inclusion. The full texts of potentially relevant studies were reviewed independently by the same 2 authors, and any studies that did not adhere to inclusion criteria were excluded. Covidence systematic review software³¹ was used to record decisions for title and abstract screening and full-text screening. Any discrepancies were resolved by discussion, and the third reviewer was consulted if necessary.

Data extraction

Data were extracted by 2 reviewers independently. A customized data extraction tool was developed in Microsoft Excel 2019 software, piloted, and used to extract data from each review based on the Joanna Briggs Institute methodology for systematic reviews.²⁹ The following data were extracted (Table 2^32–51): citation details, objectives of the included review, participant details, setting and context, number of databases searched and date range, the quality assessment tool used to appraise the included primary studies, and method of synthesis or analysis used to synthesize the evidence. For included primary studies, data on the study design, sample size, country of origin, and outcomes reported relevant to the overview review question were extracted. Statistical data on outcomes were extracted from the included reviews. If any information was missing in included reviews, primary studies were consulted, and any disagreements were discussed until consensus or by consulting the third reviewer. Primary study references are listed in Supporting Information online Table S8.

Table 2.

Characteristics and results of included systematic reviews and meta-analyses

Reference	Objectives of included review; quality assessment tool used	Participant details/setting	Databases searched (publication date range)	Included primary studies of patients with MCI: First author and date: no. of participants/study type/country of origina	Interventions reviewed (no. of primary studies that reported on intervention)	Outcomes reported (tests used)	Method of synthesis or analysis
Bird et al 202132	To systematically review evidence from RCTs investigating the acute and chronic effects of grape interventions on measures of cognition and mood in healthy participants and those with MCI Cochrane ROB Tool	Healthy participants and those with MCT	MEDLINE, the Cochrane Library, and EMBASE Database (inception–June 2020)	Krikorian 2010b: 12/RCT/USA Krikorian 2012b: 21/RCT/USA Lee 2017: 10/RCT/USA	Grape juice (2) Grape powder (1)	Memory (HVLT-R, Benton Visual Retention Test, Rey–Osterreith Complex Figure Test delayed, visuospatial–Rey–Osterreith Complex figure test copy, attention and working memory WAIS-III Letter–Number Sequencing, Montreal Cognitive Assessment and RAVLT, CVLT-II) Language or verbal skills (BNT, Letter Fluency, Category Fluency, Estimated Verbal IQ Wechsler Test of Adult Reading) Executive function (Stroop Interference, TMT– Part B, Wisconsin Card Sorting Test-64, Speed of information Processing, WAIS-III Digital Symbol, WAIS-III Symbol speed)	Qualitative analysis
Burckhardt et al 202033	To assess the effects of Souvenaid on incidence of dementia, cognition, functional performance, and safety in people with AD. Cochrane ROB Tool	Community-dwelling participants with either prodromal AD (MCI) (n = 311), mild AD dementia and mild to moderate AD dementia (n = 1097)	The specialized register of the Cochrane Dementia and Cognitive Improvement Group, MEDLINE (Ovid SP), Embase (Ovid SP), PsycINFO (Ovid SP), Web of Science (ISI Web of Science), CINAHL (EBSCOhost), Lilacs (BIREME), and clinical trials registries (database inception–June 4, 2020)	LipiDiDiet study (Soininen) 2017: 311/RCT/Finland, Germany, Netherlands, Sweden	Souvenaid (1)	Combined measure of cognition and function (CDR-SoB score); global cognition (NTB total score); memory (NTB memory domain scores); executive function (NTB executive function domain scores); incidence of dementia; AE reported	Qualitative analysis with GRADE
Butler et al 201834	To summarize the evidence on efficacy and harms of over-the-counter supplements to prevent or delay cognitive decline, MCI, or clinical Alzheimer-type dementia in adults with normal cognition or MCI but no dementia diagnosis Agency for Healthcare Research and Quality guidance	Adults with normal cognition or MCI	Ovid MEDLINE, PsycINFO, EMBASE, and the Cochrane Central Register of Controlled Trials (2009 and July 2017). Studies published before 2009 were identified by searching citations in systematic reviews and pertinent studies.	Lee 2013: 36/RCT/Malaysia Kato-Kataoka 2010: 78/RCT/Japan Peterson 2005: 516/RCT/USA Naeini 2014: 256/RCT/Iran	Omega-3 PUFAs (1) Soybean-derived phosphatidylserine (1) Vitamin E (1) Vitamins E + C (1)	Global cognitive functioning (MMSE, Hasegawa Dementia Scale, ADAS-Cog) Executive/attention/processing speed (Executive Function Attention Composite, DSST, DS Forward, DS Backward and composite battery, SDMT, number cancellation, maze tracing) Memory (memory composite, VR I, VR II, RAVLT, immediate recall and delayed recall, RBMT, HVLT, Composite Battery, presumed to be New York University Paragraph Recall Test) Visuospatial (visuospatial skills composite, Matrix Reasoning, Block Design, CLOX-1) Language (Category Fluency Test, Composite Battery (presumed to be BNT, Category Fluency) Diagnosis (possible or probable AD, CDR-SoB, AD) AE reported
Cooper et al 201335	To systematically review RCTs evaluating the effects of any intervention for MCI on cognitive, neuropsychiatric, functional, global outcomes, life quality, or incident dementia Customized criteria devised from Critical Appraisal Skills Program	Participants with MCI were recruited from clinics or clinician referrals, advertisements, care homes, the local Alzheimer’s society, preexisting research registers, a rehabilitation center, or a welfare institution.	PubMed, Web of Science, Cochrane Systematic Reviews Database, PsycINFO, CINAHL, and AMED (database inception– January 27, 2013)	Chih Chiang Chiu 2008: 23/RCT/Taiwan Sinn 2012: 50/RCT/Australia Rondanelli 2012: 25/RCT/USA Van Uffelen 2008: 152/RCT/The Netherlands De Jager 2012: 223/RCT/UK Petersen 2005: 516/RCT/USA Zhou 2007: 75/RCT/China	Omega-3 PUFA (3) B vitamins (2) Vitamin E (2)	Global cognitive functioning (CIBICplus, ADAS-Cog) Verbal functioning (letter fluency scores) Brief cognitive test performance (MMSE) Immediate memory (AVLT memory) Attention and processing speed (DSST) Executive functioning (CLOX correctly performed) Progression of MCI to AD	Qualitative and meta-analysis
Davinelli et al 202136	To evaluate the potential clinical effects of carotenoids on cognition in human adults. Cochrane ROB Tool	Mostly healthy participants and 1 study of patients with MCI (n = 4402). Age range was 45–78 years.	PubMed, Scopus, Cochrane Library, and Web of Science (database inception– July 1, 2020)	Ito 2018: 14/RCT/Japan	Carotenoids (1)	Psychomotor speed (CNSVS and ADAS-Cog subdomain scores) Processing speed (CNSVS and ADAS-Cog subdomain scores)	Individual quantitative data/meta-analysis
Dewsbury et al 202137	To evaluate the clinical evidence for the use of dietary or exogenous ketogenic agents, describe their characteristics, assess their efficacy, safety, and feasibility, and provide recommendations for future research. Cochrane ROB Tool 2.0	N = 979; MCI, n = 130. The mean age of participants was Mean pooled 68.3 ± SD pooled 7.3 y.	EBSCOHost (including CINAHL, Medline, PsychINFO), Scopus (including PubMed, EMBASE), and Web of Science (databases inception–August 21, 2020)	Fortier 2019: 52/RCT/Canada Rebello 2015: 4/RCT/USA Krikorian 2012a: 23/RCT/USA	MCT (2) Low-carbohydrate diet (1)	Global cognitive functioning (ADAS-Cog) Memory/learning (CPAL) Language Executive/attention/processing speed (TMT, DSST) AE reported	Qualitative analysis, GRADE
D’Cunha et al 201838	To analyze the recent evidence on the use of nutraceuticals and dietary supplements over the past 10 y and their associated effects on cognition in elderly individuals Criteria suggested in the Cochrane guidelines	Participants aged >65 y	PubMed, Scopus, CINAHL, and Web of Science to identify the most recent studies published in the past 10 y (from June 15, 2006, to June 14, 2016)	De Jager 2012: 266/RCT/UK Oulhaj 2016: 266/RCT/UK	B Vitamins (2)	Global cognitive functioning (MMSE, CDR, IQCODE, TICS-M) Memory (HVLT-DR, CERAD) Visuospatial (CLOX2) Informant	Qualitative analysis
Gkotzamanis and Panagiotakos 202039	To summarize and synthesize data from the latest randomized trials investigating the effect of dietary interventions on cognitive functions and the population expected to benefit most from these interventions Cochrane ROB Tool 2.0	Individuals with MCI or dementia	PubMed, Scopus, and Google Scholar (2015–2020)	Bo 2017: 86/RCT/China Zhang 2016: 240/RCT/China Boespflug 2018: 16/RCT/USA Hwang 2019: 100/RCT/Korea Remington 2015: 34/RCT/USA	Omega-3 PUFAs (2) Blueberry (1) Lactobacillus plantarum C29-fermented soybean (isoflavones) (1) Neutraceutical formulation (1)	Global cognitive functioning (BCAT, FSIQ, WAIS- RC, VLT, ACPT, DST, DRS) Working memory (Working memory task) Visuospatial domain (CLOX1)	Qualitative analysis
Jia et al 200840	The present review explored the effect not only of single vitamins, minerals, and omega-3 PUFAs on cognitive function but also their combination as might be purchased over the counter. A 10-item quality appraisal form, based on a Cochrane review	Not specified	MEDLINE, Embase, Cochrane Central Register of Controlled Trials, and CAB abstracts (database inception–September 1, 2006)	Stott 2005: 185/RCT/UK Peterson 2005: 516/RCT/USA	B vitamins (1) Vitamin E (1)	Global cognition (ADAS-cog, Global CDR, global deterioration rating, TICS-M) Processing speed (letter–digit coding test) AE reported	Meta-analysis
Li et al 202141	To evaluate the preventive efficacy of vitamin B supplements on the cognitive decline of elderly adults. Cochrane ROB Tool	Participants recruited from homes, communities, or hospitals (n = 7571). The participants of 6 studies had MCI and the other 15 studies were of elderly adults without cognitive impairment.	PubMed, Embase, The Cochrane Central Register of Controlled Trials, Web of Science, Scopus, Science Direct, PsycINFO (database inception–June 1, 2020)	Ma 2019: 120/RCT/China Kwok 2020: 279/RCT/China Ma 2017: 180/RCT/China de Jager CA 2012: 266/RCT/UK van Uffelen JG 2008: 179/RCT/Netherlands	B vitamins (5)	Global cognition (TICS-M, MMSE scores, dementia rating score, ADAS-Cog [modified], change in Global Deterioration Scale, CDR-SoB) Memory (NTB, RBMT, ADAS-cog immediate and delayed word-recall scores, and the New York University immediate and delayed paragraph-recall scores) Executive function (CLOX, NTB, Digits-Backward test, SDMT, and number-cancellation test) Information processing speed (DSST) Attention (DS-backward) Progression to AD	Meta analysis
Marx et al 201842	To examine the potential effect of resveratrol supplementation on cognitive performance and mood in adult humans. Cochrane Handbook for Systematic Reviews of Interventions checklist 2	Not specified.	Medline (via Scopus), Cumulative Index of Nursing and Allied Health Literature, Cochrane, Embase, and Proquest (database inception–June 2017)	Kobe 2017: 40/RCT/Germany	Resveratrol (1)	Verbal memory (RAVLT)	Meta-analysis, qualitative analysis
Mazereeuw et al 201243	To report the cognitive benefits of omega-3 PUFA treatment within specific neuropsychological domains across cognitively normal elderly participants, those with CIND, and patients with AD. PEDro Scale	Cognitively normal elderly adults (healthy), elderly adults with memory complaints and objective cognitive decline (CIND), and patients with dementia or AD according to standardized criteria Aged ≥50 y and without diagnosed psychiatric comorbidity	Medline, Embase, PsycInfo, Cochrane Library, AMED, and CINAHL (database inception—September 2011)	Sinn 2012: 50/RCT/Australia Kotani 2006: 21/RCT/Japan	PUFAs (2)	Global cognitive function (Memory Functioning Questionnaire, RAVLT, WAIS DS, Letter–Number Sequencing, TMT, SCWT, VFT) Composite memory, immediate recall, attention and processing Speed (RBANS [Japanese]) AE reported	Meta-analysis
McCleery et al 201844	To evaluate the effects of vitamin and mineral supplementation on cognitive function and the incidence of dementia in people with MCT Cochrane ROB Tool	Participants with a diagnosis of MCI without severe vitamin or mineral deficiencies, recruited from the community or a care home. Mean age, 66–80 y	ALOIS (the Cochrane Dementia and Cognitive Improvement Group’s specialized register), Medline, Embase, PsycINFO, CENTRAL, CINAHL, LILACs, Web of Science Core Collection, ClinicalTrials.gov, and the WHO Portal/ICTRP (database inception–January 25, 2018)	de Jager 2012: 271 (n = 223 completed the trial)/RCT/UK Fan 2017: 80 (n = 75 completed the trial)/RCT/China Eussen 2006: 119 (with MCI)/RCT/Netherlands Petersen 2005: 516 (n = 378 completed the trial)/RCT/USA Naeini 2014: 296 (n = 256 completed the trial)/RCT/Iran	B vitamins (3) Vitamin E (1) Vitamins E + C (1)	Global cognitive functioning (MMSE, ADAS-Cog [modified]), change in Global Deterioration Scale, CDR-SoB, CDR Progression to dementia Executive function (CLOX-1, NTB, Digits-Backward test, SDMT, and number–cancellation test). Processing speed (NTB) Episodic memory (ADAS-cog immediate and delayed word-recall scores and the New York University immediate and delayed paragraph-recall scores, NTB AE reported	Qualitative analysis
McGrattan et al 201845	To examine the effect of diet, either alone or in combination with lifestyle and/or cognitive strategies, on cognitive health outcomes in patients with MCI. Jadad scale and Cochrane ROB Tool	Two studies stated that participants had amnestic MCI or prodromal AD, whereas all other studies reported a diagnosis of MCI.	Ovid Medline, Embase, PsycINFO, Web of Science, and Scopus (database inception–March 2018)	Krikorian 2012a: 23/RCT/USA de Jager 2012: 266/RCT/UK Ma 2016: 180/RCT/China Lee 2013: 36/RCT/Malaysia Phillips 2015: 57/RCT/UK Bo 2017: 86/RCT/China Zhang 2017: 240/RCT/China Petersen 2005: 516/RCT/USA and Canada Desideri 2012: 90/RCT/Italy Soininen 2017 (LipiDiDiet study): 311/RCT/Finland Krikorian 2010a: 9/RCT/USA	Low carbohydrate diet (1) B vitamins (2) Omega-3 PUFA (4) Vitamin E (1) Cocoa flavanols (1) Souvenaid (1) Grape juice (1) Berries	Incident dementia and AD Memory (WAIS-RC and basic cognitive aptitude tests (information test), CERAD 10-word list learning immediate recall, delayed recall and recognition, VPAL, CVLT, TMT- part B, HVLT-R, category fluency CERAD, VR I and II, RAVLT, DS backward) Executive function and attention (category fluency, WMS-R DS total score, concept shifting test condition C (corrected for the zero trials) and LDST, HVLT-R, TMT part A and B, CLOX Psychomotor speed (DSST, TMT part A and B) Language (HVLT-R and neuropsychological measures, VFT) Visuospatial skills (WAIS-RC scores, matrix reasoning, block design) Global cognitive function (MMSE, ADAS-Cog, global CDR, CDR-SoB, composite Z score based on CERAD 10-word list learning immediate recall, CERAD 10-word delayed recall, CERAD 10-word recognition, category fluency and LDST, WAIS- RC, The Global Deterioration Scale NTB). AE reported	Qualitative analysis
Petersson and Phillipou 201646	To provide an update on the current knowledge of the effects of the MeDi on cognitive function, cognitive impairment, AD, and all-type dementia Previously customized quality assessment with additional risk-of-bias tool	No restrictions; however, most of the studies (n = 26) included participants aged >60 y, whereas the sample size ranged from 25 (in RCTs) to >17,000 participants (in cohort studies)	PubMed, Embase, CINAHL, CENTRAL, and PsycINFO (1806–May 25, 2015)	Scarmeas 2009: 1875 (n = 482 patients with MCI)/prospective cohort/USA	Mediterranean diet (1)	Cognitive function and cognitive decline (MMSE) Prevalence of cognitive impairment or incident dementia Progression of cognitive function, cognitive decline, cognitive impairment, or incident dementia (all-cause dementia or AD) AE reported	Qualitative analysis
Solfrizzi et al 201851	To provide a comprehensive systematic review of the RCTs published in the past 4 years (2014–2017) about nutritional intervention efficacy in slowing cognitive impairment progression and achieving cognitive-related outcomes in patients aged ≥60 y with late-life cognitive disorders Cochrane ROB Tool	Older patients ≥60 y with late-life cognitive disorders	PubMed, Ovid Medline, Embase, Google Scholar, Web of Science, and Scopus databases (January 1, 2014–December 31, 2017)	Remington 2015: 34/RCT/USA Cardoso 2016: 31/RCT/Brazil Lee 2017: 13/RCT/USA Fujino 2017: 328/RCT/Japan Bo 2017: 86/RCT/China	Nutraceutical formulation (1) Brazil nuts (1) Grapes (1) Plasmalogens purified from scallops (1) Omega-3 PUFA (1)	Global cognitive function (MMSE-Japanese, CERAD NTB, DRS) Memory (WMS-R scores, BCAT scores) Executive function (CLOX-1)	Qualitative analysis
Suh et al 202047	To evaluate the efficacy of vitamins without minerals or other cofactors on cognitive function of non-demented people and compared these results by geographical location and the presence of MCI. Cochrane ROB Tool	Participants were >40 years without dementia or a major neurocognitive disorder.	PubMed/MEDLINE, EMBASE, CINAHL, PsycINFO, and Cochrane Library databases (database inception to November 2019).	De Jager 2012: 271/RCT/UK Ma 2016: 180 (n = 152 completed trial)/RCT/China Van Uffelen 2008: 179 (n = 138 completed the trial)/RCT/Netherlands) Oulhaj 2014: 120/RCT/China Jiang 2013: 120/RCT/China Petersen 2005: 516 (n = 378 completed the trial)/RCT/USA and Canada Naeini 2014: 256/RCT/Iran Hu 2018: 181/RCT/China	B vitamins (5) Vitamin E (1) Vitamins E + C (1) Vitamin D (1)	Global cognition (MMSE, FSIQ, Montreal Cognitive Assessment) Memory (AVLT 1–6, HVLT-Revised with delayed recall, ADAS-cog immediate and delayed word-recall scores, and the New York University immediate and delayed paragraph-recall scores) Executive (SCWT-A task 1, 2, and 3, VFT, categorical fluency) Processing speed (DSST) Visuospatial (CLOX2, block design, picture completion, object assembly, picture arrangement) Attention (DS) AE reported	Meta-analysis
Theodore et al 202148	To present up-to date evidence regarding the association between nut intake and cognitive performance. Quality Criteria Checklist Tool, Evidence Analysis Manual, Academy of Nutrition and Dietetics.	Adults aged ≥18 y with any health condition. Total n = 43793.	Ovid MEDLINE, Scopus, CINAHL Plus, and Embase.	Cardoso 2016: 31/RCT/Brazil	Brazil nuts (1)	Global cognition (CERAD NTB, Risk of MCI) Verbal fluency, Constructional praxis, memory (CERAD NTB)	Qualitative analysis
Zhang et al 202049	To assess the scientific evidence published in the past 10 y on the effects of omega-3 PUFA intake on patients with MCI to explore whether omega-3 PUFAs have positive effects Heyland methodological quality score	N = 434	Google Scholar, Embase, and PubMed databases (2008–December 12, 2018).	(Country not reported) Baleztena 2018: 78/RCT Phillips 2015: 76/RCT Chih-Chiang 2008: 23/RCT Lee 2013: 35/RCT Hashimoto 2012: 111/RCT Rondanelli 2012: 25/RCT Mahmoudi 2014: 80/RCT	Omega-3 PUFA (7)	Global cognitive function (MMSE)	Meta-analysis
Zhu et al 202150	To evaluate the effects of probiotic and prebiotic supplementation on people with MCI and AD Cochrane ROB Tool	Both patients with AD (n = 174) and individuals with MCI (n = 446) aged 50–90 y	PubMed, Embase, Cochrane, Web of Science, and ClinicalTrials.gov (July 1, 1984–April 8, 2021)	Xiao 2020: 80/RCT/Japan Sanborn 2020: 42 (with MCI)/RCT/USA Kobayashi 2019: 121/RCT/Japan Hwang 2019: 100/RCT/Korea	Probiotics (4)	Global cognitive function (NIH Toolbox for the Assessment of Neurological and Behavioral Function, CNT [VLT, ACPT, DST], MMSE) Memory (RBANS) Visuospatial/constructional, language and attention (RBANS)	Meta-analysis

^aFor Primary study references refer to Supporting information online.

Abbreviations: ACPT, Auditory Continuous Performance Test; AD, Alzheimer's disease; ADAS-Cog, Alzheimer's Disease Assessment Scale–Cognitive Subscale; AE, adverse event; ALOIS, Dementia studies database of the Cochrane Dementia and Cognitive Improvement Group; AMED, Allied and Complementary Medicine Database; AVLT, auditory verbal learning test; BCAT, Brief Cognitive Assessment Tool; BNT, Boston Naming Test; CDR, Clinical Dementia Rating; CDR-SoB, Clinical Dementia Rating–Sum of Boxes; CERAD, Consortium to Establish a Registry for Alzheimer's Disease (cognitive battery); CFT, cognitive function test; CIBICplus, Clinician's Interview-Based Impression of Change; CINAHL, Cumulative Index to Nursing and Allied Health Literature; CIND, Cognitive Impairment Not Dementia; CLOX, clock drawing test; CVLT, California Verbal Learning Test; CNSVS, computerized neurocognitive test battery; CNT, computerized neurocognitive tests; COWAT, Controlled Oral Word Association Test; CPAL, continuous paired-associate learning; DRS, dementia rating scale; DS, digit span; DSST, Digit Symbol Substitution Test; DST, Digit Span Test; FSIQ, Full Scale Intelligence Quotient; GRADE, Grading of Recommendations; HVLT [HVLT-R], Hopkins Verbal Learning Test [Revised]; HVLT-DR, Hopkins Verbal Learning Test Delayed Recall; ICTRP, International Clinical Trials Registry Platform; IQ, intelligence quotient; IQCODE, Informant Questionnaire on Cognitive Decline in the Elderly; ISTL, International Shopping List Test; LDST, Letter–Digit Substitution Test; MCI, mild cognitive impairment; MCT, medium-chain triglyceride; MeDi, Mediterranean diet; MMSE, Mini-Mental State Examination; NIH, US National Institutes of Health; NTB, neuropsychological test battery; PEDro, Physiotherapy Evidence Database; PUFA, polyunsaturated fatty acid; RAVLT, Rey Auditory Verbal Learning Test; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; RBMT, Rivermead Behavioral Memory Test; RCT, randomized controlled trial; ROB, risk of bias; SCWT, Stroop Color and Word Test; SCWT-A, Stroop Color–Word Test Abridged; SDMT, Symbol Digit Modalities Test; TICS, Telephone Interview for Cognitive Status; TICS-M, Telephone Interview for Cognitive Status modified; TMT, trail-making test; UK, United Kingdom; USA, United States of America; VFT, Verbal Fluency Test; VLT, Verbal Learning Test; VPAL, Verbal Paired Associates Learning; VR, visual reproduction; WAIS, Wechsler Adult Intelligence Scale; WAIS-RC, Wechsler Adult Intelligence Scale, Chinese version; WMS-R, Wechsler Memory Scale, revised.

Overlap

Primary study overlap was assessed and managed using recommendations from the Cochrane Handbook for Systematic Reviews of Interventions to minimize duplication bias and overestimation of outcome effects.^51,52 All nonoverlapping systematic reviews were included or, if overlap was identified, the most comprehensive review was chosen (ie, the review that presented data from the most primary studies). If reviews were equally comprehensive, the findings of the higher-quality review was used, and if review quality was equivalent, the more recent review was included. When more than 1 review included data from the same primary study, only the data reported in the most comprehensive review were presented (Table 3).^32–51 Reviews excluded due to overlap can be found in Supporting Information Online.

Table 3.

Summary of intervention outcomes

Intervention	Most comprehensive systematic reviews that report on intervention (reference)	AMSTAR 2 rating of reviews	Primary studies reporting on interventiona	Intervention duration	Intervention summary
Mediterranean diet	Petersson and Phillipou 201646	MOD	Scarmeas (WHICAP) 2009	4 y	1 of 1 study (n = 482 participants) found a significant positive effect of moderate and high adherence to the Mediterranean diet using the MeDi score on rate of MCI to AD progression, compared with low adherence. Compared with participants in the lowest MeDi adherence tertile, those in the middle tertile had 45% less risk (HR, 0.55; 95%CI, 0.34–0.90; P = 0.01) of developing AD and those in the highest tertile had 48% less risk (HR, 0.52; 95%CI, 0.30–0.91; P = 0.02) of developing AD (for trend, HR, 0.71; 95%CI, 0.53–0.95; P for trend = 0.02). No significant association between MCI conversion to AD per unit increase in 0- to 9-point Mediterranean diet score (HR: 0.89, 95%CI, 0.78–1.02; P = 0.09)
Low-carbohydrate diet	Dewsbury et al 202137	LOW	Krikorian 2012a	6 wk	1 of 1 study (n = 23 participants) found a significant positive effect of the low-carbohydrate diet (5%–10% of calories) on memory performance using the CDR score and paired-associate learning using VPAL Task. ANCOVA F(1,20) = 6.45; P = 0.01; Cohen’s f = 0.26 (P = 0.01)
MCT	Dewsbury et al 202137	LOW	Fortier 2019 Rebello 2015	6 mo 6 mo	1 of 2 studies found a significant positive effect of MCT (30–56 g/d) on language using the BNT (52 participants). Total correct response MCT +1.0 (SD 2.2) vs placebo –1.3 (SD 2.0); P = 0.003. No other cognitive domains were significant. 1 of 2 studies had no statistical analysis (small sample size: n = 4)
B vitamins	Li et al 202141 McCleery et al 201844 Suh et al 202047 D’Cunha et al 201838 Jia et al 200840 Gkotzamanis and Panagiotakos 202039	CL MOD LOW CL CL CL	Ma 2019 Kwok 2020 Ma 2017 De Jager 2012 Van Uffelen 2008 Fan 2017 Eussen 2006 Ma 2016 Jiang 2013 Oulhaj 2016 Stott 2005 Remington 2015	6 mo 2 y 2 y 2 y 1 y 6 mo 2 y 6 mo 24 wk 2 y 1 y 1 y	Li et al41 meta-analysis of 5 studies (n = 1024), supported by an additional 5 primary studies (n = 534 participants) indicated improvements in cognition with B vitamin supplementation. Three studies (n = 583 participants) showed no effect. Li et al41: 5 studies; meta-analysis showed significant improvement (N = 1024) Meta-analysis using folic acid 0.4–5 mg/d, vitamin B12 0–0.5 mg/d and vitamin B6 0–50 mg/d (Ma 2019, Kwok 2020, Ma 2017, de Jager 2012, van Uffelen 2008) found improved global cognitive function SMD (95%CI, 0.82 (0.20–1.45; P = 0.01; I2 = 95%, test for subgroup differences, P = 0.03; information processing speed (n = 1 study), episodic memory (n = 4 studies), executive function were NS. The de Jager 2012 VITACOG study showed significant benefit of B vitamins if baseline homocysteine level was >11.3 mmol/L for MMSE (P < 0.001), episodic memory (P = 0.001), and semantic memory (P = 0.037). McCleery et al44: 1 of 2 studies showed improvement in treatment group Fan 2017 (n = 80 participants): folate 0.4 mg, B vitamins had a beneficial effect on MMSE score (unblinded) study Eussen 2006 (n = 119 participants): 1 mg vitamin B12, 1 mg vitamin B12 + 0.4 mg folic acid, or placebo for 24 wk improved memory more in the placebo group than the vitamin B12 (group z score, 0.22; 95%CI, 0.07–0.37). Neither supplementation with vitamin B12 alone nor in combination with folic acid resulted in improvements in any cognitive domains. Suh et al47: 2/2 studies significant Ma 2016 (n = 152 participants): 400 μg/d folic acid demonstrated increases in the Full-Scale IQ (P = 0.028, effect size d = 0.153), Information (P = 0.031, effect size d = 0.157), and Digit Span (P = 0.009, effect size d = 0.172) scores. Jiang 2013 (n = 120 participants): 5 mg of folic acid per day and 1.5 mg vitamin B12 per day for 6 mo. Montreal Cognitive Assessment score improved in the intervention vs control group (t test, P < 0.05) D’Cunha et al38: 1 study, significant difference Oulhaj 2016 (n = 120 participants): VITACOG study found those with highest omega-3 fatty acid levels benefitted from B vitamins (HVLTDR, P = 0.047), CDR score >0% (P = 0.043), and CDR- SoB (P = 0.04). Jia et al40: 1 study, NS findings Stott 2005 (n = 185 participants): In 3 active treatments: 2.5 mg folic acid, plus 500 µg vitamin B12, 25 mg vitamin B6, and 25 mg riboflavin, NS effect on cognitive performance as measured by the LDCT (MD, –1; 95%CI, –2.3, 1.4) and TICS (MD, –0.7; 95%CI, –1.7, 0.4). Gkotzamanis and Panagiotakos39 1 study showed improvement Remington 2015 (n = 34 participants): B vitamin formulation (folate,α-tocopherol, vitamin B12, S-adenosyl methionine, N-acetyl cysteine, and acetyl-l-carnitine) showed a positive effect on the DRS, effect size 0.79.
Vitamin D	Suh et al 202047	LOW	Hu 2018	1 y	In 1 of 1 study (n = 181 participants), there was a positive significant effect of 400 IU/d vitamin D3 on the full IQ (d = 0.70), verbal IQ (d = 0.77), performance IQ (d = 0.70), digit span (d = 1.36), vocabulary (d = 0.41), block design (d = 0.51), picture arrangement (d = 2.53); all P < 0.001.
Antioxidant vitamins (C, E)	McCleery et al 201844 Cooper et al 201335	MOD CL	Peterson 2005 Naeini 2014 Zhou 2007	3 y 1 y 1 y	Vitamin E: 1 of 2 studies reported significant finding Peterson 2005 (n = 516 participants): No significant effect of vitamin E (2000 IU/d) on progression to dementia (HR, 1.02; 95%CI, 0.74–1.4); overall cognitive functioning from MMSE, ADAS-Cog (modified) scores; clinical global impression from change in Global Deterioration Scale, CDRS-SoB; episodic memory from ADAS-Cog immediate and delayed word-recall scores, and the New York University immediate and delayed paragraph-recall scores; or executive functioning from Digits-Backward test, SDMT, and Number-Cancellation test. Zhou 2007 (n = 75 participants): 500 mg of vitamin E daily was associated with improvement in picture recognition, a secondary outcome. Vitamins E + C: 1 study, NS findings Naeini 2014 (n = 256 participants): No significant effect of 300 mg of vitamin E plus 400 mg of vitamin C per day on overall cognitive functioning determined by MMSE scores (MD, 0.23; 95%CI, –0.25 to 0.71)
Carotenoids	Davinelli et al 201836	CL	Ito 2018	3 mo	In 1 of 1 study (n = 14 participants), there was a significant positive effect of carotenoid supplementation (16 mg/d astaxanthin + sesamin) on psychomotor speed and processing speed from CNSVS and ADAS-Cog scores, both P < 0.05.
Probiotics	Zhu et al 202150	CL	Xiao 2020 Sanborn 2020 Kobayashi 2019 Hwang 2019	12-16 wk	Pooled analyses (N = 343 participants) identified a positive significant effect of probiotics on cognitive function in 4 studies (test for overall effect, Z = 3.7, P = 0.0002; I2 = 44%, P = 0.15). 1 of 2 studies supplementing with (1 × 1010 to 2 × 1010 CFU/d Bifidobacterium breve A1) found improved memory, but no significant effect on visuospatial/constructional domains, language, or attention from RBANS scores (Xiao 2020: n = 80 participants). 1 study found no significant effects for MMSE (Kobayashi 2019, 121 subjects). 1 study supplementing with Lactiplantibacillus plantarum C29 (DW2009) showed greater improvements in combined overall cognitive function and attention domain than placebo (z = 2.36, p for interaction = 0.02; z = 2.34, P for interaction = 0.02, respectively) (Hwang 2019, n = 100 participants). 1 study using L. rhamnosus found improvements in per protocol but not ITT analysis (Sanborn 2020: n = 42 participants).
Omega-3 PUFA	Zhang et al 202049 McGratten et al 201845 Cooper et al 201335 Mazereeuw et al 201243	CL LOW CL CL	Baleztena 2018 Chih-Chiang Chiu 2008 Lee 2013 Hashimoto 2012 Mahmoudi 2014 Phillips 2015 Rondanelli 2012 Bo 2017 Zhang 2017 Sinn 2011 Kotani 2006	12 mo 6 mo 12 mo 2 y 6 mo 4 mo 3 mo 6 mo 1 y 6 mo 12.8 wk	Zhang et al49: meta-analysis of 7 RCTs (n = 213 intervention; n = 221 placebo). Compared with placebo, omega-3 LC-PUFA supplements (3–24 mo) effectively improved MMSE (WMD, 0.85, 95%CI, 0.04–1.67; Z = 2.05; P = 0.04). Slight heterogeneity was detected across studies (I2 = 52%; P = 0.05) (Baleztena 2018, Chih-Chiang 2008, Lee 2013, Hashimoto 2012, Mahmoudi 2014, Phillips 2015, Rondanelli 2012) Additional studies: McGrattan et al45: 2 of 2 studies reported significant findings Bo 2017 (n = 86 participants): intervention group: 480 mg of DHA + 720 mg of EPA daily vs placebo. Working memory intervention, MD 3.32 (SD 3.45); control MD 1.38 (SD 2.66); P ≤ 0.05. Recognition memory: NS Zhang 2017 (n = 240 participants): 2 g/d DHA vs placebo showed significant improvements for short-term memory (P≤ 0.0001) and long-term memory (P ≤ 0.0001) in comparison with the placebo group. (Information test intervention mean score, 12.28 (SD 3.56); control mean score, 10.82 (SD 2.62), and digit-span intervention mean score, 13.44 (SD 3.66); control mean score, 10.25 (SD 3.42): both P ≤ 0.05. Block design test: NS). Cooper et al35: 1 study reported significant findings Sinn 2012 (n = 50 participants): in a small study compared groups receiving EPA-rich fish oil (1670 mg EPA and 160 mg DHA) and DHA-rich fish oil (1550 mg DHA and 400 mg EPA) with a placebo group. Using a linear-mixed model analysis, letter fluency scores significantly improved over 6 mo in the DHA group vs placebo LMM: t  = 2.1; P = 0.04. No other domains were significant. Mazereeuw et al43: 1 of 1 study: NS Kotani 2006 (n = 21 participants): treatment arms: 40 ARA and DHA; placebo (6 times daily) RBANS-Japanese. Duration: 12.8 wk. NS difference in any domains
Nuts (antioxidant-rich foods)	Theodore et al 202148	CL	Cardoso 2016	6 mo	1of 1 study reported significant finding Cardoso 2016 (n = 31 participants): significant effect of Brazil nut intake (estimated 288.8 μg of selenium per day) on verbal fluency (P = 0.007) and constructional praxis from CERAD subtest scores (P = 0.031), but not for overall cognition from CERAD NTB total score, or other subtests (BNT, word-list learning test, and word-list recall.
Berries (source of antioxidants)	Gkotzamanis and Panagiotakos 202039 McGrattan et al 201845	CL LOW	Boespflug 2018 Krikorian 2010a	4 mo 3 mo	1 of 2 studies reported significant finding Krikorian 2010a (n = 9 participants): positive significant effect of blueberry juice (6–9 mL of juice/kg/d) on episodic memory from VPAL score (intervention: baseline mean score, 9.3 vs week 12 mean score, 13.2); control (no detail); ANCOVA analysis intervention vs control, F1,13 = 5.58, P ≤ 0.05), but no significant effect on learning and memory from CVLT. Boespflug 2018 (n = 16 participants): NS effect of blueberries (24 g of blueberry powder) on working memory from the working memory task scores
Grape juice (source of antioxidants)	Bird et al 202132	LOW	Krikorian 2010b Krikorian 2012b	3 mo	2 of 2 studies reported significant findings Krikorian 2010 b (n = 12 participants): 6–9 mL/kg/d Concord grape juice (12 wk) improved verbal learning (CVLT scores) (intervention mean change, 3.4; control mean change, 0.0; ANCOVA intervention vs control, F(1,8) = 5.55, P = 0.04, Cohen’s f = 0.28), but NS improvement for verbal recall and spatial memory. Krikorian 2012 b (n = 21 participants): 6–9 mL/kg/d Concord grape juice (16 wk) reduced test errors (7.16 vs 5.03; adjusted means, F(1,18) = 4.53; P = 0.04; effect size Cohen’s f = 0.50). No other effects on memory and language or verbal skills MoCA and Rey Auditory Verbal Learning Test, CVLT.
Grape powder (source of antioxidants)	Bird et al 202132	LOW	Lee 2017	6 mo	1 of 1 study: NS Lee 2017 (n = 10 participants): no significant effect of grape powder consumption (72 g/d, ≈ 3 servings of fruit per day) on any cognitive domain from neuropsychological battery measures
Souvenaid	Burckhardt et al 202033	HIGH	LipiDiDiet study (Soininen 2017)	2 y	In 1 of 1 study, Souvenaid (n = 311 participants; 125 mL/d) had a significant effect on the combined measure of cognition and function from CDR-SoB scores, in modified ITT (when missing data were considered) (CDR-SoB: LME MD, −0.60, 95%CI, −1.01 to −0.19). Souvenaid probably results in a slight improvement in a combined measure of cognition and function, with this difference below estimates of meaningful change. No significant effect was found in modified ITT (missing data) on global cognition (LME MD, 0.10; 95%CI −0.04 to 0.24), memory (LME MD, 0.14; 95%CI, −0.03 to 0.30), or executive function (LME MD, −0.04; 95%CI, −0.18 to 0.10). Souvenaid probably results in little to no difference in incidence of dementia (Soininen 2017).
Resveratrol (polyphenol)	Marx et al 201842	CL	Kobe 2017	26 wk	In 1 of 1 study, there was no significant effect of resveratrol (200 mg of resveratrol and 350 mg of quercetin per day) on verbal memory from RAVLT scores (Kobe 2017: n = 40 participants).
Cocoa flavanols (polyphenol)	McGrattan et al 201845	LOW	Desideri 2012	2 mo	In 1 of 1 study, intermediate or high cocoa flavanol supplementation (520–990 mg/d) had a significant positive effect on visual attention and processing speed from the TMT, part A (high flavanol: MD, –14.3 (SD 4.2) vs intermediate flavanol: MD –8.8 (SD 3.4) vs low flavanol: MD 1.1 (SD 13.0)) and part B (high flavanol, MD, –29.2 (SD 8.0) vs intermediate flavanol, MD, –22.8 (SD 5.1) vs low flavanol, MD, 3.8 (SD 16.3); verbal functioning from the VF test (high flavanol, MD, 8.0 (SD 5.3) words/60 s); intermediate flavanol, MD 5.1 (SD 3.1) words/60 s); low flavanol, MD 1.2 (SD 2.7) words/60 s); all P ≤0.05. Global cognition Z scores significantly (P ≤ 0.05) improved in high flavanol vs low flavanol group. No difference in MMSE scores found (Desideri 2012: n = 90 participants)
Lipids (plasmalogens from scallops)	Solfrizzi et al 201851	CL	Fujino 2017	2 y	In 1 of 1 study, plasmalogens purified from scallop (1 mg/d) had no significant effect on global cognition or memory from MMSE (Japanese version) and WMS-R scores (Fujino 2017: n = 328 participants)
Soybean-derived phosphatidylserine	Butler et al 201834	LOW	Kato-Kataoka 2010	9 mo	In 1 of 1 study, soybean-derived phosphatidylserine (100–300 mg/d) had no significant effect on global cognition from MMSE scores and memory from RBMT scores) (Kato-Kataoka 2010: n = 78 participants)

^aFor Primary study references refer to Supporting information online.

Abbreviations: AD, Alzheimer's disease; ADAS-Cog, Alzheimer's Disease Assessment Scale-Cognitive Subscale; AMSTAR, A Measurement Tool to Assess Systematic Reviews; ANCOVA, analysis of covariance; ARA, arachidonic acid; BNT, Boston Naming Test; CDR, Clinical Dementia Rating; CDR-SoB, Clinical Dementia Rating Sum of Boxes; CERAD, Consortium to Establish a Registry for Alzheimer's Disease (cognitive battery); CI, confidence interval; CIBICplus, Clinician's Interview-Based Impression of Change; CL, critically low; CNSVS, computerized neurocognitive test battery; CVLT, California Verbal Learning Test; DHA, docosahexaenoic acid; DRS, dementia rating scale; EPA, eicosapentaenoic acid; HR, hazard ratio; HVLTDR, Hopkins Verbal Learning Test–Revised; IQ, intelligence quotient; ITT, intention to treat; LC, long chain; LDCT, letter-digit coding test; LME, linear mixed effects; LMM, linear mixed model; MCI, mild cognitive impairment; MCT, medium-chain triglyceride; MD, mean difference; MeDi, Mediterranean diet; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; MOD, moderate; NS, not significant; NTB, neuropsychological test battery; PUFA, polyunsaturated fatty acid; RAVLT, Rey Auditory Verbal Learning Test; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; RBMT, Rivermead Behavioral Memory Test; SD, standard deviation; RCT, randomized controlled trial; SDMT, Symbol, Digit Modalities Test; SMD, standardized mean difference; TICS/TICS-M, Telephone Interview for Cognitive Status/Telephone Interview for Cognitive Status modified; TMT, trail-making test; VF, Verbal fluency; VITACOG, Vitamins in Cognitive Impairment; VPAL, Verbal Paired-Associates Learning; WMD, weighted mean difference; WMS-R, Wechsler Memory Scale, revised.

Assessment of methodological quality of included reviews

Methodological quality of systematic reviews was assessed independently by 2 reviewers using A Measurement Tool to Assess Systematic Reviews 2 (AMSTAR 2) (Table S7 in the Supporting Information online).⁵³ Any disagreements were resolved by consensus or by consulting a third reviewer. All levels of quality were included within this review, however, were taken into account when assessing overall reliability and validity of evidence.

Data synthesis

The data was synthesized qualitatively to summarize evidence from included systematic reviews, based on type of nutritional intervention or exposure. Clinical homogeneity between studies was assessed and discussed by 2 reviewers. Final synthesized findings are presented in a ‘summary of intervention outcomes’ table (Table 3).^32–51

RESULTS

Search results

The search retrieved a total of 6677 records and 46 systematic reviews met the inclusion criteria for qualitative analysis, of which 26 reviews were not reported due to overlap of primary studies, leaving 20 studies for final analysis. The PRISMA flowchart in Figure 1 shows the selection process.

Figure 1.

Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) diagram for overview of systematic reviews of dietary patterns, foods, or nutritional supplements for individuals with mild cognitive impairment.

Description of included reviews

Included reviews were published between 2008 and 2021 and reported on 43 primary studies, all with subjects diagnosed with MCI at baseline. The relevant primary studies within the reviews were published between 2005 and 2020 with study sample sizes ranging from 4 to 1875 participants. Table 2^32–51 provides details of study characteristics for the 20 included reviews. Excluded reviews with reasons for exclusion, including those excluded due to overlap, are presented in Tables S3 and S4 in the Supporting Information online.

Interventions

The included reviews reported on the effectiveness of 18 different nutritional interventions on cognitive measures used for patients with MCI. Intervention durations ranged from 3 to 36 months and included the following dietary patterns: low-carbohydrate diet (n = 1 review); Mediterranean diet (n = 1 review); individual foods and food-derived supplements: brazil nuts (n = 1 review), grape and grape juice (n = 1 review), blueberries (n = 2 reviews), and Souvenaid (a patented drink of vitamins and minerals containing a mixture of long-chain omega-3 fatty acids; uridine; choline; vitamins B, C, and E; and selenium)⁵⁴ (n = 1 review); and the following individual nutrients: medium-chain triglycerides (MCTs) (n = 1 review), B vitamins (n = 6 reviews), D vitamins (n = 1 review), antioxidant vitamins (ie, vitamins C and E) (n = 2 reviews), carotenoids (n = 1 review), omega-3 and omega-6 PUFAs (n = 4 reviews), lipids (namely, plasmalogens purified from scallops) (n = 1 review), soybean-derived phosphatidylserine (n = 1 review), and the following phytonutrients: probiotics (n = 1 review), cocoa flavanols (n = 1 review), and resveratrol (n = 1 review). Comparator groups included normal diet, placebo, and low doses of interventions or levels of exposure.

Primary outcomes

Cognitive outcomes reported within included reviews included the rate of progression of MCI to dementia, measures of change in global cognition as well as individual cognitive subdomains, including memory, executive function, attention, processing speed, language, and visuospatial skills. Tools and tests used to measure change in cognitive function included the Mini-Mental State Examination (MMSE) scores, Alzheimer's Disease Assessment Scale–Cognitive Subscale (ADAS-Cog) scores, and various neuropsychological test batteries and specific cognitive-domain tests. All the cognitive domains and tests used within included studies are reported in Table 2 and Table S6 in the Supporting Information online.^32–51

Adverse events

Eight of the 20 included reviews reported adverse effects. Three reviews reported mild to moderate gastrointestinal disturbances (including diarrhea, nausea, abdominal discomfort, and loose stools) with intake of B vitamins,⁴⁰ omega-3 PUFA,⁴³ and MCT,³⁷ with exogenous ketones causing greater adverse events than dietary fats.³⁷ One review reported no serious adverse events.³⁴ Sleep disturbances, particularly abnormal dreams and insomnia, resulting from vitamin E³⁷ and B treatments⁴⁷ were reported in 2 reviews; however, for the vitamin E intervention, these were only reported when combined with donepezil treatments.⁴⁵ Other notable adverse events reported for omega-3 PUFA and B and E vitamins were rashes, headaches,⁴³ forgetfulness,⁴⁷ and cataract extraction.⁴⁰ Reviews also reported adverse effects in placebo groups within a vitamin E study,⁵⁵ including muscle cramps and general unwell feelings leading to discontinuation.⁴⁷ Death was reported in 2 reviews^44,47; 2 deaths occurred among patients receiving a B vitamin intervention (n = 2 of 133) and no deaths were reported for placebo groups (n = 0 of 133).⁵⁶ No difference was found between deaths in a vitamin E intervention and placebo groups (n = 5 of 257 and n = 5 of 259, respectively).⁵⁵

Methodological quality of included reviews

Thirteen reviews were assessed as being of critically low quality,^{34–36,38–41,42,43,48–51} 4 reviews were of low quality,^32,37,45,47 2 reviews were of moderate quality,^44,46 and 1 review was of high quality³³ (Table 4). The main categories within the AMSTAR 2 for which reviews scored “no” were critical domains, which ask for review methods to be established prior to conducting the review; provision of a list of excluded studies; inclusion of appropriate methods for combining statistical analyses in meta-analyses; and discussion of risk of bias in the synthesis of results. Common noncritical domains for which reviews scored “no” asked if an explanation for the study design selection was provided and if sources of funding for individual studies were reported.⁵³

Table 4.

AMSTAR2 rating of included reviews

Results for dietary patterns

Mediterranean diet

Seven reviews including 1 primary study addressed adherence to the Mediterranean diet on the risk of AD development in individuals with MCI. One moderate-quality review⁴⁶ was chosen and included 1 cohort study (n = 1875) that addressed the adherence of the Mediterranean diet on the risk of AD development in individuals with MCI (n = 482). Those in the middle tertile of adherence to the Mediterranean diet had 45% lower risk (HR, = 0.55; 95% confidence interval [CI], 0.34–0.90; P = 0.01) and those in the highest tertile had 48% lower risk (HR, = 0.52; 95% CI, 0.30–0.91; P = 0.02) of developing AD compared to the lowest tertile. No significant association was found for conversion to AD when analyzed per unit increase in a 9-point Mediterranean diet score.

Low-carbohydrate diet

Three low-quality reviews including 1 primary study addressed the effect of a low-carbohydrate diet (5%–10% of calories) on cognition, with the Dewsbury et al³⁷ study chosen as the most comprehensive. Dewsbury et al³⁷ included 1 primary study (n = 23 participants), which found a significant positive result for verbal memory performance (ANCOVA F(1,20) = 6.45; P = 0.01; Cohen’s f = 0.26 (P = 0.01).

Results for individual nutrients

Medium-chain triglycerides

Three reviews included 2 primary studies on MCTs. The report by Dewsbury et al,³⁷ a low-quality review, was chosen as the highest-quality review and included 2 primary studies, with sample sizes of 4 and 52, that addressed the impact on cognition of supplementing 30–56 g MCTs each day. A significant positive effect was found for language on the Boston Naming Test in 1 trial (n = 52)⁵⁷ using 30 g MCT/d for 6 months (MCT +1.0 (SD 2.2) vs placebo −1.3 (SD 2.0); P = 0.003.) No analysis was undertaken for the small trial with 4 participants.

B vitamins

Twelve reviews were found that discussed B vitamin interventions. Of these, 6 were chosen as the most comprehensive.^{38–41,44,47} The reviews, of critically low,^38–41 low,⁴⁷ and moderate quality rating,⁴⁴ included 12 primary studies that addressed the effect of B vitamins on cognition.

Li et al,⁴¹ whose review was of critically low quality but was the most comprehensive, conducted subgroup meta-analysis of 5 studies of patients with MCI (n = 1024); the studies used various formulations of folic acid 0.4–5 mg/d, vitamin B₁₂ 0–0.5 mg/d, and vitamin B₆ 0–50 mg/d. Results showed improved global cognitive function but with high heterogeneity (I² = 95%). Information processing speed (n = 1 study), episodic memory (n = 4 studies), and executive function (n = 3 studies) were not significantly different. An additional 5 primary studies (n = 534 participants) across 4 reviews found improvements in several cognitive domains. The 2020 low-quality review by Suh et al⁴⁷ included 2 studies that reported B vitamins were beneficial to the global cognitive function of participants with MCI, finding improvements in intelligent quotient (IQ) measures and on the Montreal Cognitive Assessment. McCleery et al⁴⁴ included 2 additional studies, 1 that found higher MMSE scores in the intervention group with supplementation with 0.4 mg of folate. Gkotzamanis and Panagiotakos³⁹ showed a nutraceutical formulation (folate, α-tocopherol, vitamin B₁₂, S-adenosyl methionine, N-acetyl cysteine and acetyl-l-carnitine) including B vitamins had a significant positive effect on the dementia rating scale (effect size 0.79). Two reviews^38,41 included primary studies from the Vitamins in Cognitive Impairment (VITACOG) trial, which indicated interaction effects. Participants with MCI and high homocysteine levels or those with highest omega-3 fatty acid concentrations had the greatest improvements in global cognition and in some cognitive domains, with B vitamin supplementation.^40,44 One additional review (ref to Jia et al 2008)⁴⁰ found no effect of B vitamin supplementation.

Vitamin D

One low-quality review⁴⁷ included 1 primary study (n = 181) that assessed 12 months supplementation of 400 IU/d vitamin D₃ on cognition. This study found a significant effect of vitamin D₃ on IQ tests compared to placebo (full IQ (d = 0.70), verbal IQ (d = 0.77), performance IQ (d = 0.70), digit span (d = 1.36), vocabulary (d = 0.41), block design (d = 0.51), picture arrangement (d = 2.53) all p < 0.001).

Vitamins C and E (antioxidant vitamins)

Ten reviews including 3 primary studies addressed the effects of vitamin E, or combined vitamin E and vitamin C on cognition. The studies by McCleery et al⁴⁴ (moderate quality) and Cooper et al³⁵ (critically low quality) were the most comprehensive. McCleery et al⁴⁴ found no significant effect of 2000 IU/d vitamin E for 36 months (n = 516 participants, with 378 completing the study) on progression to dementia, overall cognitive functioning, clinical global impression scale, episodic memory, or executive functioning. They also found the combination of 300 mg of vitamin E plus 400 mg of vitamin C per day (n = 256) had no effect on overall cognitive functioning, according to MMSE scores. Cooper et al³⁵ included an additional low-quality primary study (n = 75) showing 500 mg of vitamin E daily for 12 months was associated with improvement in the secondary outcome of picture recognition.

Carotenoids

One critically low-quality review³⁶ reported on 1 small primary study (n = 14) that explored the protective effect of carotenoids on cognition. The group supplemented with 16 mg astaxanthin and sesamin daily for 3 months had significant improvements in psychomotor speed and processing speed compared with the control group, both p < 0.05.

Polyunsaturated fatty acids

Fifteen reviews including 11 primary studies reported on the effectiveness of omega-3 PUFA supplementation on cognition. The most comprehensive reviews were those of Zhang et al,⁴⁹ McGrattan et al,⁴⁵ Cooper et al,³⁵ and Mazereeuw et al,⁴³ and these ranged from critically low^35,43,49 to low quality.⁴⁵ Zhang et al⁴⁹ analyzed 7 trials (n = 434) of omega-3 PUFAs supplements (namely, docosahexaenoic acid [DHA] 180–1300 mg, and eicosapentaenoic acid [EPA] 40–720 mg), compared with placebo for 3–24 months and found the supplementation resulted in improved MMSE scores in elders with MCI. Slight heterogeneity was detected across studies, with 4 of the 7 studies showing independent positive outcomes before meta-analysis. McGrattan et al⁴⁵ included 2 additional primary studies (n = 326) supplementing with DHA with or without EPA. Both found positive impacts on memory. Cooper et al³⁵ included 1 study (n = 50) that showed improvements in the DHA group in letter frequency but no other domains. The review by Mazereeuw et al⁴³ included 1 small Japanese study (n = 21) that compared arachidonic acid and DHA supplementation with placebo and reported no between-group differences after 12.8 weeks.

Lipid plasmalogens derived from scallops

One critically low-quality review⁵¹ reported on 1 primary study (n = 328) that explored the effects of plasmalogens derived from scallops on cognition, supplemented at 1 mg/d for 24 weeks. No significant difference was shown between the treatment and placebo groups on global cognition or memory.

Soybean-derived phosphatidylserine

One low-quality review³⁴ reported on 1 primary study (n = 78) investigating the effects of soybean-derived phosphatidylserine on cognitive outcomes, supplemented at 2 doses (100 and 300 mg daily) compared with placebo. No beneficial effect on global cognition and memory after 9 months was found.

Results for individual foods and food-derived supplements

Nuts

Two reviews of critically low quality reported on 1 primary study (n = 31) that addressed the effectiveness of Brazil nuts, as a source of selenium, on cognition. The more recent review was chosen.⁴⁸ Daily Brazil nut intake (an estimated 288.8 µg of selenium) for 6 months had a protective effect on verbal fluency and constructional praxis but not on overall cognition or other domains within the Consortium to Establish a Registry for Alzheimer's Disease neuropsychological test battery.

Berries

One critically low-quality review³⁹ and 1 low-quality review⁴⁵ reported on 2 small primary studies (n = 9 and n = 16) that investigated the effect of blueberries on cognition in MCI. One study found a significant positive effect (P ≤ 0.05) of 6–9 mL of blueberry juice per kilogram daily on episodic memory from verbal paired-associate learning scores after 12 months but no significant effect on learning or memory on California Verbal Learning Test scores. The review authors noted that the results should be interpreted with caution because of the small, potentially underpowered, sample sizes.⁴⁵ The other study found no significant effect of 24 g of blueberry powder daily for 4 months on working memory.

Grapes and grape juice

A total of 3 reviews assessed 3 studies on grape juice and grape powder. One low-quality review³² was chosen as the most comprehensive. Two small studies (n = 12 and n = 21) supplemented participants with 6–9 mL/kg grape juice daily and found improvements, one in verbal learning test scores. The second study showed reduced semantic interference on memory tasks, but the intervention did not improve the verbal learning test or other memory scores. An additional primary study (n = 10) using 72 g of grape powder (equivalent to 3 servings of fruit per day) as a source of antioxidants for 6 months found no effect on neuropsychological battery measures.

Souvenaid

Two reviews assessed 1 primary study, and the higher-quality review, by Burckhardt et al,³³ was chosen. One primary study (n = 311) found 2 years of 125 mL of Souvenaid daily improved Clinical Dementia Rating—Sum of Boxes, a combined measure of cognition and function, but this difference was reported to be small and below estimates of meaningful changes. The intervention had no significant effect on global cognition, memory, executive function, or incidence of dementia, compared with the placebo group. New data published in 2021, but not found in any included reviews show 3 years of supplementation reduced decline in the Neuropsychological Test Battery 5-item composite score (between-group difference, 0.212; 95% CI, 0.044–0.380; P = 0.014; 60% reduction in decline), Clinical Dementia Rating-Sum of Boxes (−0.90; 95% CI: −1.62 to −0.19; P = 0.014; 45% less worsening), memory (mean treatment difference 0.274, 95% CI: 0.071 to 0.477; P = 0.008; 76% reduction in decline), and brain atrophy.⁵⁸

Results for phytonutrients

Probiotics

Four reviews including 4 primary studies investigated the effectiveness of probiotics on cognition. The review by Zhu et al⁵⁰ was the most comprehensive and was of critically low quality. Pooled analyses of the 4 studies (n = 343) identified an overall positive, significant effect of probiotics on cognitive function (Z = 3.7, p = 0.0002, I² = 44%, p = 0.15). The review authors warned that the meta-analysis findings should be interpreted with caution because of differences in the probiotic strains and cognitive tests used.⁵⁰

Resveratrol

Two critically low-quality reviews reported on 1 primary study (n = 40) investigating the effect of resveratrol on cognition. The study by Marx et al⁴² was chosen because it was the more recent. There was no significant difference in verbal memory between the group supplemented with 200 mg of resveratrol and 350 mg of quercetin daily for 26 weeks, compared with the placebo group.

Cocoa flavanols

Three reviews reported on 1 primary study (n = 90) investigating the effect of cocoa flavanols on cognition. Doses were administered 3 times daily for 8 weeks. The low-quality McGrattan et al review⁴⁵ was chosen as the highest-quality review. Groups with intermediate (520 mg/d) and high levels of cocoa flavanol supplementation (990 mg/d) had significantly improved visual attention and processing speed (both P ≤ 0.05). Verbal fluency test scores significantly improved (P = 0.0001), with greater scores in the high- vs low-flavanol supplementation group. At follow-up, global cognitive function z scores only improved in the high- compared with the low-flavanol group (45 mg/d).

DISCUSSION

In this overview, we found limited evidence to conclude nutritional interventions improve or maintain cognitive function of individuals with MCI. The Mediterranean diet reduced the risk of MCI to dementia progression,¹⁰ the low-carbohydrate diet improved memory,³⁷ but only data from single studies were available. Souvenaid was the only intervention from a high-quality review³³ and although a study reported little benefit after 2 years, supplementation for 3 years has shown benefits.⁵⁸ B vitamins and omega-3 PUFA interventions were the most widely covered nutrients, and supplementation resulted in improved global cognition and individual cognition domains; however, reviews were limited in their quality.^{35,39–41,43–45,47,49} Reviews found inadequate evidence to determine the effect of supplementation of vitamins C, E,⁴⁴ and resveratrol.⁴² Low-quality reviews of single studies found vitamin D and MCTs improved cognitive domians.^37,47 For individual foods and phytonutrients, probiotics, cocoa flavanols, blueberries, and grape juice improved multiple cognition domains; however, results were from reviews of low quality, with small primary studies and/or mixed results.^39,45,50 Nonetheless, this appears to be the first overview of systematic reviews to investigate whether nutritional interventions can support cognitive function for individuals with MCI. Higher-quality primary studies and reviews are required.

Whole-diet approaches

We found evidence of 1 high-quality longitudinal primary study (n = 482 participants with MCI) that found greater adherence to the Mediterranean diet reduced the progression of MCI to dementia.⁵⁹ Many reviews have been undertaken of the Mediterranean diet and they reported benefits, but the majority included cognitively normal individuals as well as individuals with AD,^60–62 and specific outcomes for MCI are lacking. Although findings do align with the WHO risk-reduction strategy for dementia and cognitive decline, which recommends the Mediterranean diet for individuals with MCI,¹ we found little direct evidence from MCI populations. Studies should target MCI-specific populations, although this has been shown to be challenging,⁶³ and account for geographic differences, because variances between Mediterranean and non-Mediterranean countries exist.^63,64 Using validated dietary analysis tools to establish a population-specific effect is required, because the identification of a threshold of intake for improved cognition is critical, and low adherence to the Mediterranean diet has been shown to have no effect.^10,11

The DASH and MIND diets are also promising for reducing cognitive decline and dementia,^{10,13,14,16,65}; however, no reviews of these dietary patterns met our criteria for study design or they were not MCI specific. Two reviews^61,66 of dietary patterns and cognition in non-MCI populations found the Mediterranean, DASH, and MIND diets were all associated with better cognitive outcomes, with the strongest associations being for the MIND diet. Similarly, for adherence to national dietary guidelines, no suitable reviews were found. A number of primary studies using dietary guidelines have been published in different populations; all but 1 showed no impact.⁶⁷ One large study including 27,860 men and women in >40 countries found highest compared with lowest adherence to the Dietary Guidelines for Americans, measured by the modified Alternative Healthy Eating Index, lowered risk of cognitive decline regardless of baseline cognitive level,⁶⁷ suggesting benefits for people with MCI. Primary research is recommended to target these interventions to increase the range of viable dietary approaches available for reducing the progression of MCI to dementia. A 2022 review⁶⁸ of 9 prospective studies using novel neuroimaging techniques showed brain changes precede cognitive impairment and may be more sensitive than cognitive testing. In a review by Jacka et al,⁶⁹ the Mediterranean diet, Alternative Healthy Eating Index-2010, and prudent dietary pattern derived from principal components analysis were associated with improved neuroimaging markers and larger hippocampal volume in individuals without dementia. This suggests better adherence to a healthy diet may positively influence brain aging and neurodegeneration, and future studies may benefit from the use of multimodal neuroimaging techniques, as well as cognitive testing.

In the present overview, one review reported a low-carbohydrate diet improved cognition for individuals with MCI³⁷ but the primary study was small. This beneficial effect may be attributed to restricted carbohydrate intake mitigating neuroinflammation associated with impaired glucose metabolism evident in MCI,^17,18 but the finding needs confirmation in larger trials. MCT supplementation of 30 g/d to induce ketosis improved language in only 1 small trial.⁵⁷ Ketogenic agents such as supplemental MCT may be more feasible for individuals than dietary changes, although more research is required to confirm their effect and to develop agents with reduced adverse effects. The strength of evidence was poor, because it was of low quality and included a small number of primary studies with small sample sizes,³⁷ and much of the evidence for the ketogenic diet on cognition is from animal models or other population groups. Many studies, including those in the present review,³⁷ also favor the reporting of biomarkers as mechanistic and alternative measurements of cognition. For example, positive results have been found for total τ and β-amyloid-42 levels in cerebrospinal fluid of individuals eating a ketogenic diet.^37,45 A combined Mediterranean and ketogenic diet was studied in 11 people with MCI.¹⁹ Unique microbial signatures were detected in participants with MCI compared with 6 participants with normal cognition, and the Mediterranean–ketogenic diet was shown to modulate the gut microbiome and metabolites in association with improved AD biomarkers in cerebrospinal fluid. The research on the gut–brain axis and cognition is evolving quickly, but more primary studies using cognitive testing, biomarkers, and neuroimaging in human MCI populations are needed before recommendations can be made for the ketogenic diet.

Individual nutrients

Evidence from this overview suggests potentially beneficial effects of B vitamins on cognitive function of people with MCI. A meta-analysis of 5 trials⁴¹ found improved global cognition after supplementation with a variety of doses and formulations, the majority supplementing with folate and vitamin B₁₂; however, the results indicated high heterogeneity. An additional 5 of 7 primary studies also showed improvements with supplementation. Specific benefits were seen for people with MCI with high levels of homocysteine⁵⁶ and higher plasma omega-3 fatty acid concentrations⁷⁰ in reports from the VITACOG trials. These nutrient interactions indicate the importance of collecting and analyzing data based on nutritional status as dietary modification or supplementation in at-risk groups may be more likely to reduce the risk of progression to AD. High homocysteine levels are considered a risk factor for cognitive decline⁷¹ and reflect the functional status of folate and vitamins B₁₂ and B₆. Homocysteine levels can be lowered with B vitamin supplementation, yet homocysteine concentrations were either not measured or were inadequately reported in many reviews.

We found no current evidence available to confirm the synergistic effects of multiple B vitamins on cognition of individuals with MCI, because included reviews only focused on vitamins B₆, B₉, and B₁₂. One review included a nutraceutical formulation³⁹ consisting of mixed B vitamins, methionine, cysteine, and carnitine, and this combination showed favorable results on the dementia rating scale. WHO does not currently recommend B vitamins to reduce progression of MCI to dementia,¹ nor does the American Academy of Neurology.⁷² However, because these recommendations are for nutritionally replete populations, studies applying appropriate nutrient interventions for deficient individuals may strengthen the evidence base for effects of B vitamins.

PUFA interventions were conducted in 11 primary studies, and although results mostly favored the intervention, reviews reported inconsistent outcomes and were of low⁴⁵ or critically low quality.^35,43,49 The meta-analysis of 7 trials (n = 434)⁴⁹ and an additional 3 of 4 trials showed a positive impact on cognitive domains. Many studies have confirmed positive effects of fish²⁷ and fish-derived PUFAs on memory decline,⁴⁹ most likely due to improved cerebral blood flow, reduced inflammation, and changes to amyloid-β pathology.⁷³ However, to our knowledge, there is no high-quality systematic review that has examined the certainty of this evidence in MCI populations. The ApoE ε4 genotype status of participants should also be considered, because homozygous ApoE ε4 allele carriers have a higher risk of developing AD and could be resistant to protective effects of fatty acids on cognitive health.⁴⁵

We found limited evidence to conclude vitamins C, D, and E and carotenoid supplements improved cognition. A positive effect from small single studies was found for domains of processing speed from carotenoids, attention from vitamin D, and with 1 of 2 studies on vitamin E treatment showing impact on picture recognition and no significant effects for combined vitamin C and E treatments.⁷⁴

Vitamins play a wide range of roles regulating the central nervous system and, therefore, could affect cognitive decline.⁴⁴ However, limited literature is available to confirm findings in MCI populations, with evidence focusing on improvements to dementia biomarkers rather than outcomes of cognitive tests and impacts of vitamin deficiencies rather than effects of supplementation.⁴⁴ Studies using vitamin supplements may be affected by reduced bioavailability, increased susceptibility of nutrients to heating and drying, nutrient aging or lifespan and degradation in transport and storage, as well as potential interactions with other nutrients.⁷⁵ The inclusion of predominantly well-nourished participants in some reviews may also limit outcomes, because individuals with low vitamin intakes or poor-quality diets may respond differently. Studies should focus on increasing consistency between dose and preparation of supplements, be specifically targeted to MCI populations, assess nutrient intake, and investigate groups with poor nutritional status.^76,77

Individual foods and food-derived supplements

One high-quality review,³³ including the 2017 LipiDiDiet study of 2 years’ supplementation of the Souvenaid nutritional drink, found no improvement in cognition alone but did find improvements in Clinical Dementia Rating—Sum of Boxes, a combined measure of cognition and function. The difference was small, however, and no reduction in progression to dementia was found.³³ Less-than-predicted cognitive decline was seen in the control group and the trial was continued. After 3 years of Souvenaid supplementation, the 2021 report of the LipiDiDiet study⁵⁸ indicated there were significant reductions in decline in the Neuropsychological Test Battery 5-item composite score (between-group difference 0.212 (95% CI: 0.044 to 0.380); p = 0.014; 60% reduction in decline), Clinical Dementia Rating-Sum of Boxes (−0.90, 95% CI: 1.62 to 0.19; p = 0.014; 45% less worsening), memory (0.274, 95% CI: 0.071 to 0.477; p = 0.008; 76% reduction in decline), as well as reductions in measures of brain atrophy. Results of the 3-year study indicate that early initiation in prodromal AD and adequate treatment duration are important for benefits. Many of the trials in this overview were of short duration; as shown in the LipDiDiet study, trial duration must be long enough to see significant changes in cognition of the placebo group.

High-antioxidant foods have been shown to improve cognition.⁷⁸ In this overview, we found a positive effect of nuts and blueberries on cognition; however, evidence was from reviews of low quality, reported from small studies, and detected in specific cognitive tests only. Including a wider range of cognitive markers could provide insight into the mechanisms underlying the progression of dementia. For example, Gkotzamanis and Panagiotakos³⁹ used functional magnetic resonance imaging to show increased stimulation of certain brain regions from blueberry interventions. In this overview, we found no cognitive advantage of reconstituted whole-grape powder consumption alone, contrary to their expected effects, most notably from their high antioxidant content. A positive effect on memory was found for the more concentrated grape juice, potentially due to higher resveratrol content in the intervention dose.⁴⁵ However, when resveratrol was isolated, no beneficial effect on memory was demonstrated.⁴² These results align with those of other recent primary studies investigating resveratrol treatments for patients with MCI⁷⁹ or healthy older adults.³⁴ Although this indicates individual high-antioxidant foods may have limited effects, these foods are still recommended for the prevention of cognitive decline,⁷⁸ especially as part of a healthy antioxidant-rich diet. Studies are recommended to deduce whether combinations of high-antioxidant foods exert synergistic effects on cognitive outcomes in patients with MCI.

There was only low-quality evidence for the effectiveness of phytonutrient supplements on cognition in patients with MCI. Positive results were reported for probiotics, carotenoids, and cocoa flavanols, but findings were from small or heterogeneous studies and were reported in low-⁴⁹ to critically low-quality reviews.³⁶ Therefore, whether probiotics, resveratrol, cocoa flavanols, or carotenoids improve or do not improve cognitive function in MCI remains uncertain.

Although the present review explored the effects of nutrition alone on cognition in the MCI population, multidomain trials that combine nutrition with other forms of therapy have been undertaken. The large Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability (FINGER) trial (n = 1190)⁸⁰ of at-risk individuals found yearly between-group differences in neuropsychological test battery scores of 0.022 (95%CI, 0.002–0.042; P = 0.030) favoring the multidomain interventions of nutrition, exercise, brain training, and vascular risk monitoring. The researchers found 72% of participants were actively involved in all 4 intervention domains and that the program can provide substantial societal benefits.⁸¹ This contrasts with a recent Cochrane review by Hafdi et al⁸² that found no evidence of multidomain intervention benefits for preventing incident dementia. Hafdi et al⁸² did find small improvements in cognitive function in the group receiving the multi-domain intervention, although authors note this was strongest in trials with cognitive training and may therefore be linked to a learning effect.

Adverse effects were found for B and E vitamins, MCTs, and omega-3 PUFA interventions, with mainly mild to moderate symptoms reported, including gastrointestinal and sleep disturbances.^37,40,43,47 Only 8 of the 20 reviews reported adverse events, indicating underreporting. The nocebo effect was also possible, because adverse effects were seen in both intervention and placebo groups. This finding conforms to those of other reviews using nutritional interventions where the nocebo effect exerted a small to moderate effect on study outcomes.⁸³ Death was reported in 2 reviews,^44,47 but the reviews’ findings may have been confounded by the older age of study participants with unreported comorbidities. Supplements are a multimillion-dollar industry but are not without risk⁸⁴ and studies should improve reporting of adverse events so individuals can be appropriately informed when adopting a nutritional intervention to support their cognitive function.⁸⁵

Limitations

Reviews included in the present overview were of generally low quality, particularly in critical domains including establishing research methods prior to conducting the review, provision of a list of excluded studies, and discussion of risk of bias. The heterogeneity within reviews may have also introduced bias. The strength of the evidence presented in included reviews was also limited by the minimal data reported on adverse effects, and the small, potentially underpowered sample sizes in primary studies. Furthermore, our review only presented data from the most comprehensive reviews to minimize overlap, and bias introduced from overestimating effect sizes. Although this meant that some higher-quality reviews were excluded, this strategy for managing overlap was necessary to manage the large volume of data. Last, our search was limited to systematic reviews of RCTs and cohorts. Because narrative reviews were excluded, additional individual studies may have been missed. We acknowledge that only reviewing evidence from study designs of the highest standard can be reductionistic and that valuable evidence can come from observational studies as well as from emerging primary research that has not been reviewed.

Future research

Because whole dietary approaches are more likely to improve cognition than supplementation with specific nutrients,¹ the neuroprotective effects of whole diets and dietary patterns, including the MIND diet⁶ and the ketogenic diet,¹⁸ should be prioritized in future research. Increasing the duration of interventions should also be a focus of primary research, because some studies were only conducted for 6 to 8 weeks.⁸⁶ Future trials should consider the likely risk-related population conversion rates of MCI to dementia of approximately 10% per year⁸⁷ and ensure the follow-up time is able to show clinically relevant cognitive or neuroimaging change in the placebo group. The adequate time period for studies is difficult to determine; nevertheless, including studies lasting at least 2–3 years may be more likely to show cognitive changes.⁸⁸ Establishing baseline dietary intake and nutritional status are important for researchers to consider, because greater cognitive protection has been seen in some groups (eg, those with high omega-3 fatty acid and homocysteine levels).^71,76,89 More common genetic polymorphisms (eg, ApoE and the dihydrofolate reductase 19 base pair deletion)⁹⁰ and medications (aspirin)⁹¹ should also be considered as having a potential impact on treatment, particularly for B vitamins.⁸⁹ Additionally, it may be prudent to include biomarkers and neuroimaging as outcomes, in addition to cognitive testing, to better detect short-term effects of interventions. Interventions and reviews should focus specifically on patients with MCI, and if combined with other populations, should provide adequate participant numbers and clearly report statistical analysis of participants with MCI.

To our knowledge, this overview is the first to use systematic review methodology to consolidate evidence on dietary patterns, nutrients, and foods to support cognitive function for individuals with MCI; however, our most salient finding was the lack of high-quality evidence on the topic. Our results suggest that B vitamins, omega-3 fatty acids, the Mediterranean diet, and Souvenaid are potentially protective of cognition in patients with MCI. The uncertainty of the evidence on remaining interventions for vitamins D, C, E, and carotenoids prevented further recommendations for this population, although it seems reasonable and safe to recommend the Mediterranean diet, based on its associated overall health benefits. Although dementia remains a critical public health problem, more high-quality research is warranted to provide more tailored nutritional guidelines for individuals with MCI.

Supporting Information

The following Supporting Information is available through the online version of this article at the publisher’s website.

Table S1 PRISMA checklist

Table S2 PRISMA 2020 abstract checklist

Table S3 Excluded studies

Table S4 Excluded studies due to overlap

Table S5 Search Strategies

Table S6 Cognitive Tests Included in Review

Table S7 AMSTAR 2 review questions

Table S8 Primary study references