Table 1.

Characteristics of studies included in qualitative analysis of association of Mediterranean diet with Alzheimer’s disease and mild cognitive impairment.

First Author, Year and Country	Sample Size	Mean Age (Years)	Study Desing	Diet Measures: FFQ → Diet Score (Scale)/Pattern	Other Tests and Measures	Results
Ballarini et al., 2021 Germany [30]	N = 512	69.5	CS Groups: CN (N = 169) MCI (N = 81) SCD (N = 209) AD (N = 53)	EPIC-FFQ → MDs (1–9) -Low MDA ( 0–3) -Medium MDA (4–5) -High MDA (6–8)	-MRI volumetry -NPT (ADAS-COG; WMS; CERAD; SDMT; FNT; FCSRT; VF; BNT; TMT, CDT) -Cerebrospinal fluid (Aβ42/40 ratio, pTau181)	Higher MDA related to larger mediotemporal gray matter volume (p < 0.05), better memory (p = 0.038), and less amyloid (p = 0.008) and pTau181 pathology (p = 0.004)
Karstens et al., 2019 USA [10]	N = 82	68.8	CS Groups: ND (N = 82)	BFFQ 2005 → MDs (0–55) -Median split: -Low MDA (N = 39) -High MDA (N = 43)	-MRI volumetry -NPT (CVLT-II, TMT, WAIS-IV digit symbol coding, WAIS-IV letter number sequencing subtest, WTAR, MMSE, BDI, BAI) -MFSRP -BMI	High adherence to MD is related with better learning and memory in NPT (p = 0.007) and with larger dentate gyrus volumes compared with low MDA (p = 0.03). MDA is not related with information processing or executive functioning in NPT and neither with white matter hyperintensity.
Walters et al., 2018 USA [1]	N = 70	49	L for 3 years Groups: CN (N = 70)	HWSQFFQ → MDs (0–9) - Continuous variable	-MRI volumetry -FDG-PET -PiB-PET -NPT (WAIS digit symbol substitution, WAIS vocabulary, MMSE, paragraph recall, paired associates recall, object naming, design tests) - Vascular risk measures (BMI, blood pressure, plasma cholesterol/HDL ratio, plasma homocysteine QUICKI) -APO-E G -MLTAQ -Intellectual activity through life interview	Lower adherence to MD is related with faster decline in FDG-PET (p < 0.05). Adherence to MD is not related with NPT, PiB-PET or MRI measures. Exercise and intellectual activity are not related with changes in AD biomarkers or NPT.
Calil et al., 2018 Brazil [11]	N = 96	75.2	CS Groups: CN (N = 36) MCI (N = 30) AD (N = 30)	FFQ → MDs (0–55) -Tertiles: -Low MDA -Middle MDA -High MDA	-NPT (MMSE, BCSB, VF, CDT, GDS) -BMI	Higher adherence to MD is related with higher MMSE and BCSB Learning scores in CN group (p < 0.05). No associations are found between other NPT outcomes and MDs. No associations are found between dietary patterns and NPT outcomes in MCI or AD participants.
Hill et al., 2018 Australia [31]	N = 115	70	CS Groups: Women of the Women’s Health Ageing Project (N = 115)	DQESv2 → -High fat pattern (N = 24) -MD pattern (N = 31) -Junk food pattern (N = 24) -Low fat pattern (N = 35)	-F18F-PET -NPT (CERAD) -BMI -APO-E G	Adherence to junk food pattern was related with higher F18F-PET measures (p = 0.03). Other dietary patterns are not related with F18-PET.
Rainey-Smith et al., 2018 Australia [14]	N = 77	71.5	L for 3 years Groups: CN (N = 77)	CCVFFQ → MDs (0–9) -Continuous variable	-PiB-PET -APO-E G -BMI -NPT (MMSE)	Higher MDs is related with lower PiB-PET measures (p = 0.007).
Vassilaki et al., 2018 USA [8]	N = 278	77.7	CS Groups: CN (N = 278)	MB1995RHHFFQ → MDs (0–9) -Continuous variable	-PiB-PET -BMI -NPT	Higher MDs (p = 0.012), vegetable consumption (p = 0.002), Vitamin A (p = 0.003) and β-carotene intakes (p = 0.005) and moderate alcohol consumption (p = 0.03) are related with lower PiB-PET measures.
Berti et al., 2018 USA [7]	N = 70	50	CS and L for 3 years Groups: CN (N = 70)	HFFQ → MDs (0–9) -Median split: -Low MDA (N = 36) -High MDA (N = 34)	-NPT (GDS, HDRS, MMSE, CDR, WAIS digit symbol substitution, paired associates recall, paragraph recall, design tests, object naming) -MRI volumetry -FDG-PET -PiB-PET -Vascular risk measures (BMI, blood pressure, plasma cholesterol, triglycerides, plasma homocysteine, fasting glucose, hip-to-waist ratio, QUICKI, fasting glucose)	Low MDA is related with lower FDG-PET measures and higher PiB-PET measures compared with high MDA at baseline (p < 0.001). Low MDA is related with greater FDG-PET declines and PiB-PET increases compared with high MDA longitudinally (p < 0.001). No relation is observed between MDA and MRI volumes.
Hill et al., 2018 Australia [32]	N = 111	69.7	CS Groups: Women of the Women’s Health Ageing Project (N = 111)	DQESv2→ MDs (0–18 ) -Continuous variable -Tertiles: -Low MDA (N = 56) -Middle MDA (N = 32) -High MDA (N= 23)	-NPT -MRI -F18F-PET -IPAQ -BMI -IPAQ-E -AACVRs	There is no significant relation between MDA and F18F-PET measures.
Mosconi et al., 2018 USA [3]	N = 116	50	CS Groups: CN (N = 116)	BFFQ y HFFQ → MDs -Continuous variable	-MRI volumetry -NPT (CDR, GDetS, HDRS, memory, WAIS digit symbol substitution, WAIS vocabulary) -Intellectual activity through life 25-item interview -Vascular risk measures (BMI, blood pressure, plasma cholesterol, plasma homocysteine, QUICKI) -Baecke and Minnesota leisure time physical activity questionnaires	Higher MDA and higher insulin sensitivity are both significant related with higher MRI volumetry measures (p < 0.08). No other lifestyle and vascular risk variables are significant related with MRI volumetry measures. Higher MRI volumes are significant related with better cognitive performance. Intellectual enrichment is related with better cognition (p < 0.01)
Merrill et al., 2016 USA [33]	N = 44	62.6	CS Groups: SMI (N = 24) MCI (N = 20)	5 points Likert scale of Mediterranean-type diet -Often adherence to MD -Rarely adherence to MD	-FDDNP-PET -BMI -IPAQ-E -NPT (MMSE, HRSD, HRSA) -MRI	MCI group with above normal BMI have higher FDDNP-PET binding than MCI group with normal BMI (p = 0.02). Higher physical activity is related with lower FDDNP-PET binding in MCI group (p = 0.04) but not in SMI group. Higher consume of MD is related with lower FDDNP-PET binding in both groups (p = 0.04)
Morris et al., 2015 USA [27]	N = 923	58–98 (Range)	L for 4.5 years Groups: ND (N = 923)	HFFQ → MDs (0–55) -Continuous variables -Tertiles: -Low MDA -Middle MDA -High MDA	-NPT (CESDS) -APO-E G -BMI -Cognitively stimulating activities self-reported -Physical activity time spent self-reported	High adherence to MD have significant lower rates of AD incidence than low adherence to it (p for trend = 0.006).
Olsson et al., 2015 Sweeden [34]	N = 1038	71	L for 12 years Groups: Men CN (N = 1038)	Seven days food record prepared by Swedish National Food Administration→ modified MDs(0–8) -Continuous variables -Tertiles: -Low MDA -Middle MDA -High MDA	-NPT (MMSE) -APO-E G -Vascular risk measures (BMI, blood pressure, plasma cholesterol, HDL and LDL cholesterol, serum triglycerides, insulin sensitivity) -CRP levels	Higher MDA is potentially- not significantly- related with lower risk of developing all-type cognitive impairment (but not with AD or all-type dementia risk).
Matthews et al., 2014 USA [35]	N = 45	54	CS Groups: CN (N = 45)	HFFQ →MDs (0–9) -Median split: -Low MDA -High MDA	-MLTAQ -MRI volumetry -FDG-PET -PiB-PET -NPT (CDR, MMSE, HDRS, MHIS, WAIS vocabulary, WAIS digit symbol substitution, paired associated recall, paragraph recall, designs, object naming) -APO-E G -Vascular risk measures (BMI, HTWR, blood pressure, plasma cholesterol, HDL and LDL cholesterol, blood glucose, serum triglycerides, insulin sensitivity)	Lower physical activity is related with higher PiB-PET measures, lower FDG-PET measures and reduced MRI measures than higher physical activity (p < 0.001). Low MDA is related with higher PiB-PET measures, lower FDG-PET measures and reduced MRI measures than high adherence (p < 0.001). Significant interactions effects between physical activity and MDA are seen in FDG-PET measures (p = 0.003).
Mosconi et al., 2014 USA [20]	N = 52	54	CS Groups: CN (N = 52)	HFFQ → MDs (0–9) -Continuous variable -Median split: -Low MDA -High MDA	-MRI volumetry -NPT (CDR, MMSE, HDRS, MHIS, GDetS, WAIS vocabulary, WAIS digit symbol substitution, paired associated recall, paragraph recall, designs, object naming) -Vascular risk measures (BMI, HTWR, blood pressure, plasma cholesterol, HDL and LDL cholesterol, blood glucose, serum triglycerides, plasma homocysteine, insulin sensitivity) -APO-E G	High MDA is related with greater MRI measures in left hemisphere AD-vulnerable regions compared with low MDA (p = 0.026). MDA is not related with cognitive performance.
Gardener et al., 2012 Australia [16]	N = 970	71.72	CS Groups: CN (N = 723) MCI (N = 98) AD (N = 149)	CCVFFQ →MDs (0–9) -Continuous variable	-NPT (MMSE, LM II, D-KEFS Verbal Fluency, CVLT II Long Delay) -BMI -APO-E G	AD group has lower MDA than CN group (p < 0.001). MCI group has lower MDA than CN group (p < 0.05). MDs is related with changes in MMSE over 18 months period in CN group (p < 0.05).
Gu et al., 2010 USA [36]	N = 1219	76.7	CS and L for 4 years Groups: ND (N = 1219)	HFFQ → MDs (0–9) -Continuous variable -Tertiles: -Low MDA -Middle MDA -High MDA	-NPT (memory, language, processing speed and visual-spatial ability) -High sensitivity CRP plasma levels -Fasting insulin serum levels -Serum total adiponectin levels -APO-E G -BMI -Modified CIC	Higher MDA is related with lower levels of hsCRP (p = 0.003). Higher MDA is not related with levels of fasting insulin or total adiponectin. Higher MDA is related with lower risk of developing AD (p for trend = 0.04). Association between MDA and AD risk of incidence did not seem to be mediated by high sensitivity CRP, fasting insulin or total adiponectin levels.
Scarmeas et al., 2009 USA [17]	N = 1880	77.2	L for 14 years Groups: ND (N = 1880)	HFFQ →MDs (0–9) -Continuous variable -Tertiles: -Low MDA -Middle MDA -High MDA -Median split: -Low MDA -Middle MDA -High MDA	-NPT (memory, orientation, abstract reasoning, language, visual-spatial abilities, CDR) -GLTEQ -BMI -CIC -APO-E G	Middle MDA compared with low MDA reduces AD risk with HR= 0.98 (95% CI 0.72–1.33), while high MDA compared with low MDA reduces AD risk with HR = 0.6 (95% CI 0.42–0.87), (p for trend= 0.008). Some physical activity compared with no physical activity reduces AD risk with HR = 0.75 (95% CI 0.54–1.04), while much physical activity compared with no physical activity reduces AD risk with HR = 0.67 (95% CI 0.47–0.95) (p for trend = 0.03). Much physical activity and high MDA compared with no physical activity and low MDA reduces AD risk with HR = 0.65 (95% CI 0.44–0.96) (p for trend = 0.03).
Scarmeas et al., 2009 USA [13]	N = 1875	76.9	L for 10 years Groups: CN (N = 1393) MCI (N = 482)	HFFQ →MDs (0–9) -Continuous variable -Tertiles: -Low MDA -Middle MDA -High MDA	-NPT (memory, orientation, abstract reasoning, language, visual-spatial abilities, CDR) -BMI -APO-E G	Middle MDA compared with low MDA is borderline related with lower risk of developing MCI (p = 0.24). High MDA compared with low MDA is related with lower risk of developing MCI (p = 0.05). Middle MDA compared with low MDA is related with lower risk of developing AD from MCI (p = 0.01). High MDA compared with low MDA is related with lower risk of developing AD from MCI (p = 0.02).
Feart et al., 2009 France [37]	N = 1410	75.9	L for 5 years Groups: CN (N = 1410) AD (N = 66)	HFFQ →MDs (0–9) -Continuous variable -Tertiles: -Low MDA -Middle MDA -High MDA	-NPT (MMSE, IST, BVRT, FCSRT) -APO-E G	Higher MDA, was associated with slower MMSE cognitive decline but not with other cognitive tests this relationship was attenuated when adjusting for stroke. Higher MDA was not associated with risk for incident dementia.
Scarmeas et al., 2006 USA [9]	N = 1984	76.3	Nested Case-control Groups: ND (N = 1790) AD (N = 194)	HFFQ →MDs (0–9) -Continuous variable -Tertiles: -Low MDA -Middle MDA -High MDA	-NPT (memory, orientation, abstract reasoning, language, visual-spatial abilities, CDR) -APO-E G -BMI -Modified CIC -Vascular risk measures (BMI, plasma cholesterol, HDL and LDL cholesterol, blood glucose, serum triglycerides, plasma homocysteine, insulin sensitivity)	Higher MDA is related with lower risk of AD (p < 0.001). High and middle MDA are related with lower risk of AD compared with low MDA (p for trend < 0.001). Vascular variables do not change de magnitude of the association.
Scarmeas et al., 2006 USA [38]	N = 2258	77.2	L for 10 years Groups: ND (N = 2258)	HFFQ →MDs (0–9 pts.) -Continuous variable -Tertiles: -Low MDA -Middle MDA -High MDA	-NPT (memory, orientation, abstract reasoning, language, visual-spatial abilities, CDR) -APO-E G -BMI	Higher MDA is related with lower risk of AD incidence (p = 0.003). High and middle MDA are related with lower risk of AD incidence compared with low MDA (p for trend = 0.007).

ADAS-COG: Alzheimer’s Disease Assessment Scale; AACVRs: Australian Absolute Cardiovascular Risk Score; AD: Alzheimer Disease; APO-E G: Apolipoprotein E genotype; BAI: Beck Anxiety Inventory; BCSB: Brief Cognitive Screening Battery; BDI: Beck Depression Inventory; BFFQ: Block Food Frequency Questionnaire; BMI: Body Mass Index; BNT: Boston naming test; CCVFFQ: Cancer Council of Victoria Food Frequency Questionnaire; CDR: Clinical Dementia Rating; CDT: Clock Drawing Test; CERAD: Consortium to Establish a Registry for Alzheimer's Disease; CESDS: Center for Epidemiological Studies-Depression Scale; BVRT: Benton Visual Retention Test; CID: Charlson Index of Comorbidity; CN: Cognitively Normal; CRP: C-Reactive Protein; CS: cross-sectional; CVLT-II: Californian Verbal Learning Test Second Edition; D-KEFS: Delis-Kaplan Executive Function System; DQESv2: Dietary Questionnaire for Epidemiological Studies Version 2; FDDNP-PET: 2-(1-(6-[(2-[F-18]fluoroethyl)(methyl)amino]-2naphthyl)ethylidene)malononitrile Positron Emission Tomography; FCSRT Free and Cued Selective Reminding Test; FDG-PET: Fluoro-Deoxy-Glucose Positron Emission Tomography; EPIC: German adaptation of the semiquantitative European Prospective Investigation of Cancer. FFQ: Food Frequency Questionnaire; F18F-PET: Fluoro 18 Florbetaben Positron Emission Tomography; GDetS: Global Deterioration Scale; GDS: Geriatric Depression Scale; GLTEQ: Godin Leisure Time Exercise Questionnaire; GVT: Greek vocabulary Test; GVLT: Greek Verbal Learning test; HARS: Hamilton Anxiety Rating Scale; HDRS: Hamilton Depression Rating Scale; HFFQ: Harvard Food Frequency Questionnaire; hsCRP: high-sensitivity C reactive protein; HTWR: Hip to Waist Ratio; IST: Isaacs Set Test; IPAQ-E: International Physical Activity Questionnaire modified for older adults; JLO: Judgment of Line Orientation L: Longitudinal; LM II: Logical Memory II; MFSRP: modified Framingham Stroke Risk Profile; MB1995RHHFFQ: Modified Block 1995 Revision of the Health Habits Food Frequency Questionnaire; MCI: Mild Cognitive Impairment; MCG: medical college of Georgia complex figure test; MD: Mediterranean Diet; MDA: Mediterranean Diet Adherence; MDs: MedDiet score; MHIS: Modified Hachinski Ischemia Scale; MLTAQ: Minnesota Leisure Time Activity; MMSE: Mini-Mental State Examination; MRI: Magnetic Resonance Imaging; ND: non-demented; NPT: Neuropsychological tests; PiB-PET: Pittsburgh Compound B Positron Emission Tomography; pts.: points; QUICKI: Quantitative Insulin Sensitivity Check Index; SCD: Subjective Cognitive Decline; SDMT Symbol digit modalities test; SMI: Subjective Memory Impairment; TMT: Trail Making Test; VF: Verbal Fluency; WAIS-IV: Wechsler Adult Intelligence Scale IV; WTAR: Wechsler Test of Adult Reading; WMS: Wechsler Memory Scale.