Genetic instrumental variable analysis reveals multivariate dietary patterns beneficial to patients with Alzheimer's disease

Abstract

Background

Dietary patterns (DPs) are important in the modulation of the pathogenesis of Alzheimer's disease (AD) and progression of neurodegeneration. However, literature reports are inconsistent in studies examining how DPs benefit AD patients.

Objective

To assess potential causal links between diets and AD risk occurrence.

Methods

Based on comprehensive genome-wide association studies data, potential links between diets and AD risk were assessed by conventional and advanced Mendelian randomization and sensitivity analyses using single nucleotide polymorphisms as genetic instruments.

Results

Protein (95% CI: −0.280 to −0.250), fat (95% CI: −0.091 to −0.042), and carbohydrates (95% CI: −0.048 to −0.021) displayed protective roles in AD. The role of sugar, one component of carbohydrates, was ambiguous in development of AD. 24 DPs derived from daily diet habits were negative in the association with AD and 24 DPs of daily diet habits positively associated with AD. Many single diet factors such as beer, decaffeinated coffee, non-oily fish, lamb/mutton, and biscuit cereal were found to display harmful or protective roles in AD development and progress. However, their roles were varied in DPs of individual diet habits and showed significant positive or negative loadings to AD risks dependent on DPs, not dependent on single diet factors themselves.

Conclusions

Significantly detrimental or protective loadings to AD risks are dependent on DPs of individuals, not dependent on the single diet factor itself. Data here suggests that, for principles, daily DPs should balance protein, fat, including saturated fatty acids and unsaturated fatty acids, and carbohydrates to prevent AD development and progress.

Keywords

Alzheimer's disease beverages diet pattern and habits macronutrients single food intake

Introduction

Alzheimer's disease (AD) is one of the most expensive, insidious, and burdening diseases and the main cause of dementia in elderly populations. It is estimated that there are more than 400 million people with AD dementia, prodromal AD, and preclinical AD worldwide, projected to triple in 2050.^1,2 In the 85-year-old population, the prevalence of AD is 3 times higher than that of clinically defined AD.³ A US-based study evaluating survival after a dementia diagnosis in almost 60,000 individuals reported survival times of 3–4 years.^3,4 Despite extensive efforts to find a cure and prevention approaches, AD still is the fifth-leading cause of death among adults aged 65 or older and remains incurable.⁵

The strongest risk factors for AD are advanced age and carrying at least one apolipoprotein E type 4 (APOE ε4) allele.^1,6 Moreover, women are more likely to develop AD than are men, especially after the age of 80 years.² Alterations in neurons, microglia, astroglia, neuro-inflammation, vessels, aging, and dysfunction of the glymphatic system act upstream or in parallel to the pathological developments of AD.⁷ In addition, cardiovascular risk factors and lifestyles have been associated with an increased AD risk.^8,9 However, evidence suggests that vascular risk factors do not increase the risk of AD pathology as measured by cerebrospinal fluid biomarkers or PET.¹⁰ Lifestyle factors, such as the level of education, physical activity, sleep, diet, smoking, and alcohol consumption are identified as being involved in the AD pathological progresses.^11,12 Unhealthy lifestyles such as smoking, alcohol consumption, lack of sleep, or physical inactivity have been associated with an increased AD risk.^8,9 Given that there is currently no treatment on AD, the significance of exploring lifestyle risk factors for improving cognitive health of elderly people are conducted. Dietary and lifestyle modifications may prevent the development of AD patients.¹² A healthy diet may have a beneficial effect on human cognitive function and reduce the risk of developing AD.^12,13 Dietary patterns (DPs) for the elderly are often advised to individuals for reduction of AD risks and to patients with AD for slowing down progresses of the disease. However, literature data are inconsistent in studies examining these DPs benefit to AD patients.^14,15 Since dietary habits are significantly heritable,^16–20 another question remains is how easy individuals shift dietary habits that have been established during their lifetime to recommended diet patterns. Thus, a better DP that can be more easily modified for individuals is required for people to reduce AD risks and to prevent AD development and progress.

Mendelian randomization (MR) uses germline genetic variants strongly associated with exposures of interest as a natural experiment to investigate the causal relations between potentially modifiable risk factors and health outcomes in observational data.²¹ In one measure, MR analysis does not provide evidence to support the causal genetic relationships between dietary factors and AD risks.²² In contrast, MR analysis identifies that increased intake of dried fruits and oily fish are causal protective factors for AD and the processed meat, poultry, and beef intaking positively associated with AD. In addition, a higher relative intake of protein can causally reduce the risk of AD, and a higher relative intake of fat may be protective against AD.^23–25 However, one evidence suggests that dried fruit intake is a positive causality to AD.²⁶ No evidence showed that AD was associated with carbohydrate, sugar intake, and fresh fruit intake.^25,26 Taking them together, the results from MR analysis are controversy. Thus, systematically analysis of diet patterns in the AD risks is required to clarify the relationship between diet factors and AD risks. In present work, based on comprehensive genome-wide association studies (GWAS) data,^16–20^,27,28 we assessed potential causal links between diets and AD risk occurrence, thereby offering valuable insights into future clinical practices.

Methods

Genetic instrument selection for exposure

The independent single nucleotide polymorphisms (SNPs) of selected exposure factors as indicated in the paper were identified from corresponding meta-analyses of GWAS (https://prism.northwestern.edu/; https://www.kp4cd.org/) in European ancestries.^16–20 The GWAS data of single food intake, beverages, and multivariate DPs are generated from over 500,000 individuals aged 37–73 years living throughout the UK.^16,20 The heritability and GWAS analysis of single food intake, analyzed as curated single food intake quantitative traits, and of principal component-derived DPs of real-world dietary habits using Food Frequency Questionnaire data in up to 500,000 Europeans from UK Biobank (UKB) as described.^16,20 Briefly, dietary intake was collected using a 24 h recall questionnaire (Oxford WebQ) in a subset of UK Biobank participants. Total bitter beverages included coffee, tea, grapefruit juice, and bitter tasting alcoholic beverages (beer/cider, red wine, and liquor) and total sweet beverages included sugar sweetened beverages, artificially sweetened beverages, pure non-grapefruit juices, flavored milk, and hot chocolate were defined, subsequently conducted GWAS of ‘beverage-sets’. Self-reported consumption of alcohol, coffee, and tea was also collected from all participants at baseline using a touchscreen questionnaire. Dietary habits derived 85 curated single food intake (FI) quantitative traits (FI-QTs) from Food Frequency Questionnaire (FFQ) data of Europeans from UKB, using 35 nested and complementary questions. The UKB FFQ consists of quantitative continuous variables, ordinal non-quantitative variables depending on overall daily/weekly frequency, food types, or foods never eaten. All 85 single FI dietary phenotypes were then adjusted for age in months and sex, followed by inverse rank normal transformation on continuous FI-QTs. For individuals with repeated FFQ responses, both the dietary variable and the age in months covariate were averaged over all repeated measures. DPs as described by principal component (PC) analysis are derived from food groups with the highest factor loads for each DP are such as processed meat, vegetables, fish, beans, bread, refined grains, whole grains, spread, fruit, wine, milk, et al.²⁰ Phenotype correlation between 170 dietary habits is estimated and 85 DPs, as described as PC1 to PC 85, are then derived.²⁰ The food names or trademarks are used as ones presented in the literature. The GWAS data of habitual coffee consumption data are from 91,462 coffee consumers of European ancestry.¹⁷ The GWAS data of fat, protein, and carbohydrate intake from 268,922 individuals of European ancestry and The GWAS data of sugar, one component of carbohydrates, intake are from 235,391 individuals of European ancestry.¹⁸ The GWAS of carbohydrates includes intake from all saccharides. The GWAS of sugar includes intake from mono- and disaccharides not only added and refined sugar (found in for instance, sugar-sweetened beverages and candy), but also natural sugars, found mainly in fruit and dairy products.¹⁸ The GWAS data of tobacco and alcohol were used to control the measurements and were not shown in the present paper.¹⁹ Names and markers of single diet factors as described in the original literature are not changed in line with general names of single foods or beverages in the present paper.^16,20

Selection of SNPs as instrumental variables

At the genome-wide significance threshold of p < 5 × 10⁻⁸, except pure fruit juice that was at the genome-wide significance threshold of p < 5 × 10⁻⁷, we extracted those SNPs significantly associated with selected exposure factors. For each SNP, the R² value indicates the effect size of a genetic variant and quantifies how much of the total variation in the trait. The F-statistic tests the null hypothesis that the SNP has no effect on the trait. The R² and F statistics of each SNP were calculated using the formula: R² = 2 × MAF × (1 − MAF) × d², F statistic = R² × (N − 2)/(1 − R²). To refine our selection, we excluded any SNPs that exhibited linkage disequilibrium. Set parameters of linkage disequilibrium set to R²< 0.001 within a 10,000 kb window to ensure the selected SNPs are not correlated with each other. SNPs with an F-statistic less than 10 are considered weak instruments and are removed to avoid are instrument bias. The SNPs of all exposure factors with an F statistic were above 29 in the present measurements. We then extracted the corresponding SNP-outcome effect estimates from the outcome GWAS and harmonized the datasets to ensure allele alignment. To ensure reliability, we extracted more than 15 SNPs robustly associated with a given exposure factor as instrumental variables (IVs) to examine the relationship with AD.

Outcome data sources and meta-analysis

We analyzed the GWAS summary statistics pertaining to AD, harnessing data from the prominent, publicly available cohorts, including 455,258 individuals first (71,880 cases, 383,378 controls) and then including 1,126,563 individuals with 12,968 additional cases and 488,616 additional controls.^27,28 AD GWAS summary statistics were obtained from the website (https://pgc.unc.edu/). We eliminated gene variants with low confidence and focused solely on those with a minor allele frequency greater than 0.01 for our meta-analysis. A z-score meta-analysis of AD summary statistics was conducted between samples using METAL. After rigorous quality control, we narrowed down the variants from both cohorts to only those that met our standards. The resulting set of 13,354,030 variants was then used for the meta-analysis.

Conventional Mendelian randomization and sensitivity analyses

The effect of exposure factors on AD outcomes was estimated using the random-effects inverse variance weighted (IVW), MR-Egger, weighted median, simple mode, and weighted modes.²⁹ The IVW method was applied to obtain the primary outcome. MR-Egger, weighted median, simple mode, and weighted mode methods were performed to verify the causal effects. We harmonized SNPs to ensure that the effect estimates of each SNP on each trait and the risk of AD corresponded to the same alleles. To assess the heterogeneity across each SNP, we employed Cochran's Q statistics and Leave-One-Out Sensitivity Analysis.³⁰ Additionally, we utilized the MR-Egger intercept test and the Mendelian Randomization Pleiotropy RESidual Sum and Outlier (MR-PRESSO) to detect and correct potential horizontal pleiotropy.³¹ All these analyses were carried out using TwoSample MR (version 0.6.14), Mendelian Randomization (0.10.0) and MR-PRESSO (version 1.0) in R (version 4.5.0). P-value <0.05 was considered as the significant level.

Analysis of advanced Mendelian randomization techniques

Contamination mixture methods (CMM) in MR was performed to account for invalid IVs due to horizontal pleiotropy, as such conditions, genetic variants affect the outcome through pathways other than the exposure factors.³² CMM was used to estimate the proportion of invalid instruments and adjust causal estimates accordingly.

Robust Adjusted Profile Score (MR-RAPS): MR-RAPS is a powerful and robust MR method, particularly useful when dealing with weak instruments and measurement errors as described.³³ When the biases in MR analysis were observed, MR-RAPS was performed to measure error and weak instruments.

Bayesian Model Averaging in Mendelian Randomization (MR-BMA): When the biases in MR analysis were observed, MR-BMR also was performed to incorporate different MR methods (e.g., IVW, MR-Egger, weighted median) or varying sets of genetic instruments, accounting for model uncertainty to avoid over-reliance on a single model, reducing bias from misspecification as described.^34,35

Mendelian Randomization using Least Absolute Shrinkage and Selection Operator (MR-LASSO): When the biases in MR analysis were observed, we also performed MR-LASSO that was used to select strong genetic instruments in high-dimensional settings as described.³⁶ Combined with other advanced MR techniques to handle pleiotropy and selection bias, MR-LASSO was used to be sure of the reliability in MR applications.

Analysis of advanced MR techniques carried out using MR (0.10.0) in R (version 4.5.0). P-value <0.05 was considered as the significant level.

Results

Macronutrients associated with AD risk

To test for the causal effect of macronutrients on the AD risk, 487, 308, 256 and 265 SNPs strongly associated with carbohydrate that includes intake from all saccharides, sugar that includes intake from mono- and disaccharides, fat and protein, respectively, were used as genetic instruments (Table s1, s2) to determine their effects on AD risks. We first performed conventional MR analyses including MR Egger, Simple mode, Weighted mode, Weighted median, and Inverse variance weighted assays. The MR results showed carbohydrates were positively associated with AD. Fat and protein were negatively associated with AD. In contrast, Simple mode assay showed sugar was negative with AD. The other assays showed sugar was positively associated with AD. Then we used Cochran's Q test for heterogeneity of the genetic instruments and Egger's Regression Intercept Test to assess directional pleiotropy of genetic instruments. The results showed that genetic instruments with carbohydrate, sugar, fat and protein were high heterogeneity and horizontal pleiotropy and indicated that the reliability of results from conventional MR methods was substantially weakened and unreliable. Thus, we performed MR-PRESSO and Leave-One-Out Sensitivity Analysis to detect and correct horizontal pleiotropy and outlier genetic instruments. After removal of the outliers, the genetic instruments associated with carbohydrate, sugar, fat and protein also showed high heterogeneity and directional pleiotropy. To be sure of the reliability of results from MR assays, we employed advanced MR techniques, including CMM, MR-RAPS, BMA-MR, and MR-LASSO to analyze the genetic instruments. The results confirmed that fat and protein were strongly negative to associate with AD (Figure 1, Supplemental Tables 2 and 3), which are consistent with previous observation.²⁵ Carbohydrates were negatively associated with AD (Figure 1, Supplemental Tables 1 and 3). In contrast, the results from MR-LASSO and CMM showed that sugar was negatively associated with AD. The BMA-MR and MR-RAPS showed that sugar was positively associated with AD (Figure 1, Supplemental Tables 2 and 3). The examinations suggest that sugar might display positive or negative roles during development of AD. The examinations do not clarify the roles of sugar in the development of AD.

Figure 1.

Forest plot of macronutrients in associations with risks of AD. Beta value, p value and 95% CIs of the significant macronutrients from the valid, reliable and robust methods were reported. CI: confidence interval.

Beverages and single food intake associated with AD risk

Since macronutrients are obtained from single foods and beverages, we performed MR analysis using GWAS data of beverages to determine whether any beverage consumption was associated with AD risks. Alcohol consumption exists extensively, and acts as one of the environmental risk factors of AD. Indeed, the results of genetic instruments tightly associated with overall alcohol consumption showed that alcohol consumption was a risk factor for AD (Figure 2, Supplemental Tables 4 and 5). Next, we determined which alcohol consumption increased risks of AD (Figure 2, Supplemental Tables 4 and 5). The results showed that bitter alcohol (red wine, liquor/spirits, beer/cider) was well negatively associated with AD. So did beer and cider, both were negatively associated with AD. If people only drank during meals, alcohol consumption was also negatively associated with AD (Figure 2, Supplemental Tables 4 and 5). Champagne and white wine significantly increased risks of AD. The examinations suggest that alcohol consumption is not always harmful in the development of AD. It seems to act as protective roles in the development of AD in some ways.

Figure 2.

Forest plot of beverages in associations with risks of AD. Beta value, p value and 95% CIs of the significant beverages from the valid, reliable and robust methods were reported. CI: confidence interval.

We identified that milk, including full cream milk, dairy based milk, semi−skimmed milk, skimmed milk, and soy milk, were negatively associated with AD (Figure 2, Supplemental Tables 4 and 5). Among them, full cream milk showed most significantly protective roles in AD development. Bitter drinks, including ground and instance coffee and tea, and water also showed greatly negative roles during developments of AD (Figure 2, Supplemental Tables 4 and 5). In contrast, decaffeinated coffee showed harmful roles in AD patients. Examinations of genetic instruments also showed that pure fruit juice and sugar sweetened beverages were positively associated with AD (Figure 2, Supplemental Tables 4 and 5).

Available comprehensive GWAS data of single food intake gave us opportunities to measure the relationship between single food intake and AD development. We performed conventional MR analyses and advanced MR techniques and identified that 16 single intaking foods were negatively associated with risks of AD and two foods displayed risk factors of AD (Figure 3, Supplemental Tables 6 and 7). The highly protective food of AD was non-oily fish. We identified that overall meat intake, lamb/ mutton, and poultry were negative with risks of AD, controversy to the previous observations.^23–25 In addition, oil based spreads, olive oil-based spread, cooked vegetables, cornflakes and Frosties, cereal, butter, oily fish, fresh fruit, whole grain, dried fruit, white bread, and raw vegetables displayed negative roles in development of AD. Among cereal, biscuit cereal, and oat cereal were positively associated with AD. Thus, taken together, we show that many kinds of beverages and single foods are negatively or positively associated with the development of AD.

Figure 3.

Forest plot of single diet factors in associations with risks of AD. Beta value, p value, OR and 95% CIs of the significant diet factors from the valid, reliable and robust methods were reported. CI: confidence interval; OR: odds ratio value; SNPs: single-nucleotide polymorphisms.

Dietary patterns associated with AD

Next, we performed conventional MR analyses, outlier detection and correction of MR-PRESSO and Leave-One-Out Sensitivity Analysis, and advanced MR techniques to analyze genetic instruments tightly associated with DPs using the GWAS data generated by Cole et al.²⁰ The 85 PCs of DPs were analyzed. The results showed that 48 DPs were associated with AD risks. Among them, 24 DPs were negatively associated with AD risks, and 24 DPs positively associated with AD risks (Supplemental Tables 8 and 9, Supplemental Figure 1). For instance, one DP, PC1, which is described as containing foods like Western (i.e., red and processed meats, sugary drinks, high-fat products, and refined grains) and prudent dietary factors (characterized by fruits, vegetables, fish, whole grains, and low-fat dairy),²⁰ was significantly negative in the association with AD risks. The top two DPs negatively associated with AD risks were PC55 and PC42 and the top two DPs positively associated with AD risks were PC22 and PC23. For instance, PC1 has significant positive loading foods including wholemeal/wholegrain bread consumption, increased fruit and vegetable intake, increased oily fish intake, and increased water intake and significant negative loading foods include white bread consumption, butter and oil spread consumption, increased processed meat intake, and consumption of milk with higher fat content.²⁰ PC55 has significant positive loading foods including bowls of cereal, muesli, cheese intake, and negative loading foods including adding salt to food, dairy-based milk, milk, semi-skimmed milk, skimmed milk, and non-oily fish intake. PC22 positive loads oat cereal, cooked vegetables, never eat dairy, never eat sugar, oily fish intake, tea, water and negatively loads bowls of cereal, cornflakes/Frosties, champagne/white wine, cheese intake, ground coffee, fortwine, fresh fruit, hot drinks, and overall processed meat intake.²⁰

Next, we compared diet patterns positively associated with AD risks and diet patterns negatively associated with AD risks to identify the food loadings. The results showed that while single food or beverage intake showed significantly negative or positive in the association of AD, it did not always substantially affect the same ways to contribute to the heritability of AD risks in diet patterns (Supplemental Table 10, Figure 4). For instance, the single tea intake was significantly negative in the association with AD risk. However, in DPs PC22 and PC23, tea was positively associated with AD risks. In such DPs, tea drinks were the risk factor for AD patients. In diet patterns such as PC42 negatively associated with AD risk, tea drinks acted as a protective factor in AD development. The spirit drink as a single diet factor did not show any connection with AD risks; however, in some diet patterns, it showed a risk or a protective factor for AD (Supplemental Table 10). We identified that there were no diet factors consistently shown to be protective or harmful factors measured as a single factor and in diet patterns. For instance, champagne and white wine were positively associated with AD risks measured as a single factor but acted as protective factors in some diet patterns. Cooked vegetables, lamb/mutton, and non-oily fish as measured as a single factor showed protective roles in AD; however, in some cases they acted as harmful factors in diet patterns during AD development. Thus, our determination indicates that diet factors play complex roles in AD development when individual diet patterns are measured. Diet factors playing harmful or protective roles in AD are dependent on individual diet patterns.

Figure 4.

The correlations among AD and single diet factors in dietary patterns. The heatmaps show the top two dietary patterns positively and negatively associated with AD. The correlation coefficients of single diet factors among all the dietary patterns were derived from Cole et al.²⁰

Discussion

We show here significantly detrimental or protective loadings to AD risks dependent on diet patterns of individuals, not dependent on the single diet factor itself. For instance, Mediterranean diets link to the homeostatic formation of hippocampal neurons and carry benefiting effects upon their nutritional status and cognitive function have been suggested to form the basis of nutritional recommendations for people with AD.^37,38 Mediterranean diet is characterized by high consumption of vegetables, fruits, nuts, legumes, unrefined grains, and low consumption of meat and dairy products, the total fat content may be moderate or high, ranging from 30–40% of the total daily energy requirement.^12,37,39 It also characterizes high consumption of olive oil and fish and low meat consumption and the increased consumption of red wine with meals. Another recommended DP is DASH diet. It is intended to prevent high blood pressure and reduces fat intake, consumption of red meat, sweets, and sugar-containing beverages and is rich in nutrients such as dietary fiber, calcium, magnesium, and potassium, with a low sodium intake.^11,39,40 The third recommended DP for people with AD is the MIND diet, a combination of the Mediterranean diet and the DASH diet.⁴¹ MIND diet contraindicates red meat, butter, cheese, sweets, fried products, and fast food and is proposed to reduce the risk of developing cognitive disorders, including AD and is based on plant-based foods, emphasizing consuming large amounts of leafy greens, nuts, and blueberries. The three DPs have been shown to slow down the decline in cognitive function in general and in individual domains of cognition.⁴² However, studies show how the impact of the Mediterranean diet on AD is unclear.^14,15 DASH diet is shown not associated with better cognitive outcomes.^15,43 MIND diet adherence has been found positively and significantly correlated with baseline cognition but not significantly associated with slower cognitive decline over a 6-year period.⁴² Those observations suggest that current recommendatory diet patterns are not well beneficial to AD patients and to elderly for AD prevention.

For instance, we identify a DP, PC1,²⁰ which is similar to western DPs characterized by a high intake of processed meat, red meat, butter, high-fat dairy products, eggs, and refined grains and prudent DP characterized by a high intake of vegetables, fruit, legumes, whole grains, and fish and other seafood,^20,44 is negatively associated with AD risks, suggesting that individuals who have the dietary habits are not necessary to shift their dietary habits. We also identify single diet factors positively contribute to a DP, PC23,²⁰ that is highly positive in the association with AD are oat cereal, muesli cereal, decaffeinated coffee, cooked vegetables, lamb/mutton intake, and tea, negatively slices of bread, bowls of cereal, bran cereal, cornflakes/wFrosties cereal, fresh fruit, hot drinks, olive oil spread, low fat spread, any oil-based spread, and water. Thus, individuals who have such dietary habits might decrease intake oat cereal, muesli cereal, decaffeinated coffee, cooked vegetables, lamb/mutton intake, and tea and increase the consumption of slices of bread, bowls of cereal, bran cereal, cornflakes/Frosties cereal, fresh fruit, hot drinks, olive oil spread, low fat spread, oil-based spread, and water. Further measurements indicate that many identified single diet factors, which are negatively in association with AD risks, positively contribute to diet patterns that are highly positive in the association with AD risks, indicating again that significantly detrimental or protective loadings to AD risks dependent on diet patterns of individuals, not dependent on the single diet factor itself. In summary, our measurements indicate not all diet factors and diet patterns are involved in AD risks in human daily life. The dietary roles in AD development and progression are mostly dependent on multivariate DPs of individuals, but not on single diet factors including single foods and beverages. In general, any single diet factor including foods and beverages in DPs provide protein, fat and carbohydrate required for human body. Consistently, we identify that protein, fat, and carbohydrates are all negatively associated with AD risks and might be required to prevent the development of AD.

Presently, few works address strictly the question whether supply of protein in the diet can benefit the health of people who have AD. Yeh et al. show that replacing 5% of energy in the diet from protein, with an equivalent amount of energy from carbohydrates, is associated with a higher probability of cognitive decline. When 5% of energy replacement from vegetable protein with animal protein can reduce the risk of cognitive decline. In consistent with the observations, plant protein intake had a higher risk of cognitive decline.^45–48 Higher animal protein intake was associated with a lower risk of cognitive decline.^45–51 In addition, one cohort shows the potentially protective impact of high dietary protein intake on brain β-amyloid burden in older adults.⁵² In Japanese aged over 60, protein intake significant decreases rates of AD disability-adjusted life year (AD-DALY) across all sex and age groups.⁵³ The AD-DALY rates were high with additional plant protein and low with increased animal protein intake in males in their 60 s and the AD-DALY rates were high with adding animal protein and low with additional plant protein intake in males aged over 70. In females, the estimated AD-DALY rates decreased as animal or plant protein intake increased.⁵³ The observations support our finding that protein is required to prevent AD. However, data from 204 countries suggest that red and white meat supply may collectively influence dementia risk.⁵⁴ In an AD mouse model, periodic protein restriction reduced cognitive deficits as well as phosphorylation of the tau protein.⁵⁵ Branched-chain amino acids (BCAA; leucine, isoleucine, and valine) are essential amino acids, comprising approximately 20% of protein intake. BCAAs have the potential to serve as a biomarker for dementia and AD.⁵⁶ In an AD animal model, BCAA supplementation combined with a high-fat diet resulted in a significant mortality rate and strongly increased tau neuropathology. Reducing BCAA in high fat diet did not result in any premature death, improved cognitive performance.⁵⁷ Those observations suggest that nutritional strategies aimed at reducing AD, while maintaining protein intake, should reduce the BCAA content in the diet supply.⁵⁷ Thus, a protein supply with quantity and quality should be paid attention when dealing with the AD prevention and the health of people with AD.

Currently, several works in literature directly address the requirements of fat for the clinical management of people with AD. The works regarding the beneficial or harmful effects of a high-fat diet on AD show controversy. Unsaturated fatty acids have been found to act as a protective effect against AD, while saturated fatty acids (SFA) have higher risks of the disease.^39,58 A high supply of total fat and SFA in the form of milk and spreads at midlife has been identified with poorer global cognitive function and prospective memory and increase risks of mild cognitive impairment.⁵⁹ In addition, unsaturated fatty acid consumption is better for cognitive function and sensory memory.^59,60 In contrast, studies show that DPs containing increased short-chain SFA and middle-chain SFA specifically lauric acid intake are less likely to suffer from cognitive impairment.⁶¹ Peripheral saturated long-chain fatty acids are associated with increased risk of progressing from mild cognitive impairment to AD.⁶² The consumption of medium-chain triglycerides also has been shown to improve cognitive function in mild cognitive impairment.⁶³ Increased fat intake and decreased carbohydrates display better global cognition among older adults at risk for dementia.⁶⁴ In our present observations, fat might display a protective role in the development of AD. Our findings by genetic instruments provide evidence to support the fact that a DP with fat intake may be recommended for people with AD.

At moment, carbohydrate intake in people with AD is not precisely defined. Many observations indicate that people with impaired glycemia progress more rapidly from mild cognitive impairment to AD.⁶⁵ People with type II diabetes carrying increased serum glucose have a higher AD risk.⁶⁶ Carbohydrate intake is positively associated with the cerebral loading of amyloid-β in people, and sugar consumption was associated with cognitive declines.^67–69 A high-carbohydrate diet associated with increased blood glucose levels might affect human cognitive function negatively.⁷⁰ The observations suggest that a high-glycemic diet might lead to an increased AD risk. Excessive sugar consumption among older adults showed a notable association with poor cognitive functions.^71,72 Cognitive impairment has been found when intake approximates levels of sugar consumption in people. Increased absolute sugar intake is significantly associated with a higher risk of all-cause dementia and AD.⁷³ In women population, excessive total sugar intake is significantly associated with AD risks and lactose has a stronger impact on AD risk.⁷⁴ In contrast, in animal models, deficiency of dietary fiber can lead to cognitive disorders and to structural changes in the hippocampus and disturbances in the gut microbiota of mice, subsequently impair cognition.^75–77 Higher dietary fiber intake is associated with improved specific components of cognitive function in older adults aged 60 and older.⁷⁸ Our current findings show that carbohydrate intake from all saccharides might act as a protective factor of AD.

In contrast, our results show that MR analyses yielded conflicting evidence regarding the causal effect of sugar intake that includes intake from mono- and disaccharides on AD risk. Specifically, MR-LASSO and CMM indicated a negative association (sugar intake potentially protective against AD), whereas BMA-MR and MR-RAPS suggested a positive association (sugar intake increasing AD risk). This inconsistency likely arises from differences in how each method handles horizontal pleiotropy, weak instruments, and model uncertainty. MR-LASSO selects a sparse set of genetic instruments, which may exclude pleiotropic variants and inadvertently remove instruments that capture harmful sugar-related pathways. In contrast, BMA-MR averages across many models, potentially giving weight to pleiotropic instruments that drive a positive signal. CMM is designed to correct for certain patterns of horizontal pleiotropy but may be sensitive to violations of its underlying assumptions, while MR-RAPS is robust to weak instruments and outliers yet can produce biased estimates when the distribution of pleiotropy is non-normal. In addition, sugar intake is a complex exposure often measured by self-report, and genetic instruments may reflect different biological subtypes (e.g., fructose versus glucose) or correlate with other dietary and metabolic factors (e.g., total energy intake, obesity, inflammation) that influence AD in opposite directions. Given these inconsistencies, we caution against interpreting any single MR estimate as definitive. Instead, our findings highlight that current genetic instruments for sugar intake do not provide robust, unambiguous support for either a protective or harmful causal effect on AD. Thus, further examinations, such as well-controlled observational studies, animal models, and randomized trials of sugar-restricted diets, are required to draw clinical or public health conclusions.

It is limitations that the results presented here are correlated with AD risks and development and are not solid evidence to demonstrate dietary factors can be used as AD risk, prevention and treatment approaches. Clinical trials are required to address how dietary factors are involved in AD patients in future. Anyway, our data, combining the literatures, suggests that, for principles, the DPs of individual daily habits should balance protein, better animal protein, fat, including saturated fatty acids and unsaturated fatty acids, carbohydrates, including a small amount of sugar for individuals potentially to prevent AD development and progress (Figure 5). Individuals who have very particular tastes in food might be required to address synergistic nutrition of their daily diet patterns to balance protein, fat, and carbohydrates.

Figure 5.

Principle requirements of the dietary patterns of individual daily habits for prevention of AD development and progress.

Supplemental Material

sj-xlsx-1-alr-10.1177_25424823261451159 - Supplemental material for Genetic instrumental variable analysis reveals multivariate dietary patterns beneficial to patients with Alzheimer's disease

Supplemental material, sj-xlsx-1-alr-10.1177_25424823261451159 for Genetic instrumental variable analysis reveals multivariate dietary patterns beneficial to patients with Alzheimer's disease by Hong Xu, Yutong Nie, Siying Li, Junman Ye, Qiaorong Huang, Wentong Meng, Xue Li and Xianming Mo in Journal of Alzheimer's Disease Reports

Supplemental Material

sj-docx-2-alr-10.1177_25424823261451159 - Supplemental material for Genetic instrumental variable analysis reveals multivariate dietary patterns beneficial to patients with Alzheimer's disease

Supplemental material, sj-docx-2-alr-10.1177_25424823261451159 for Genetic instrumental variable analysis reveals multivariate dietary patterns beneficial to patients with Alzheimer's disease by Hong Xu, Yutong Nie, Siying Li, Junman Ye, Qiaorong Huang, Wentong Meng, Xue Li and Xianming Mo in Journal of Alzheimer's Disease Reports

Footnotes

Acknowledgements

The authors thank all the investigators for sharing the data used in this study. The sharing data were obtained from Zhong et al., Hum Mol Genet (2019), https://prism.northwestern.edu/; Coffee and Caffeine Genetics Consortium, Mol Psychiatry (2015); Meddens et al., Mol Psychiatry (2021), https://www.thessgac.org/; Liu et al., Nat Genet (2019), https://www.kp4cd.org/; Cole et at., Nat Commun (2020), https://www.kp4cd.org/ and Jansen et al., Nat Genet (2019), Wightman et al., Nat Genet (2021), . We also thank FinnGen, UKBB, Psychiatric Genomics Consortium, and Social Science Genetic Association Consortium for data resources.

ORCID iD

Xianming Mo

Ethical considerations

This study utilized publicly available de-identified data from participant studies and therefore no separate ethics approval was required in this study.

Consent to participate

This study utilized publicly available de-identified data from participant studies and therefore no separate consent of participation was required in this study.

Consent for publication

Not applicable

Author contribution(s)

Hong Xu: Conceptualization; Data curation; Formal analysis; Methodology; Writing – review & editing.

Siying Li: Methodology; Visualization.

Junman Ye: Methodology; Validation.

Qiaorong Huang: Resources.

Wentong Meng: Methodology; Project administration.

Xue Li: Methodology; Visualization.

Xianming Mo: Conceptualization; Formal analysis; Project administration; Resources; Supervision; Validation; Writing – original draft; Writing – review & editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the 1.3.5 project for disciplines of excellence of West China Hospital (ZYGD23026).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The results of this study are included in this published article and its supplementary information files. All analyses in this study were conducted using R version 4.5.0 () with standard functions and packages as detailed in the Methods section. No custom analytic code was generated. No unique research materials were generated in this study.

Supplemental material

Supplemental material for this article is available online.

References

Gustavsson

Norton

Fast

, et al. Global estimates on the number of persons across the Alzheimer's disease continuum. Alzheimers Dement 2023; 19: 658–670.

Scheltens

De Strooper

Kivipelto

, et al. Alzheimer's disease. Lancet 2021; 397: 1577–1590.

Manly

Jones

Langa

, et al. Estimating the prevalence of dementia and mild cognitive impairment in the US: the 2016 health and retirement study harmonized cognitive assessment protocol project. JAMA Neurol 2022; 79: 1242–1249.

Gillis

Montenigro

Nejati

, et al. Estimating prevalence of early Alzheimer's disease in the United States, accounting for racial and ethnic diversity. Alzheimers Dement 2023; 19: 1841–1848.

Alzheimer's Association. 2024 Alzheimer's disease facts and figures. Alzheimers Dement 2024; 20: 3708–3821.

Corder

Saunders

Strittmatter

, et al. Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer's disease in late onset families. Science 1993; 261: 921–923.

Morris

Spillantini

Sue

, et al. The pathogenesis of Parkinson's disease. Lancet 2024; 403: 293–304.

Saeed

Lopez

Cohen

, et al. Cardiovascular disease and Alzheimer's disease: the heart-brain axis. J Am Heart Assoc 2023; 12: e030780.

Ornish

Madison

Kivipelto

, et al. Effects of intensive lifestyle changes on the progression of mild cognitive impairment or early dementia due to Alzheimer's disease: a randomized, controlled clinical trial. Alzheimers Res Ther 2024; 16: 122.

10.

Ricci

Martorana

Bonomi

, et al. Effect of vascular risk factors on blood-brain barrier and cerebrospinal fluid biomarkers along the Alzheimer's disease continuum: a retrospective observational study. J Alzheimers Dis 2024; 97: 599–607.

11.

Dhana

Franco

Ritz

, et al. Healthy lifestyle and life expectancy with and without Alzheimer's dementia: population based cohort study. Br Med J 2022; 377: e068390.

12.

Dominguez

Veronese

Parisi

, et al. Mediterranean Diet and lifestyle in persons with mild to moderate Alzheimer's disease. Nutrients 2024; 16: 3421.

13.

Phillips

MCL

Deprez

Mortimer

GMN

, et al. Randomized crossover trial of a modified ketogenic diet in Alzheimer's disease. Alzheimers Res Ther 2021; 13: 51.

14.

Grant

Blake

. Diet's role in modifying risk of Alzheimer's disease: history and present understanding. J Alzheimers Dis 2023; 96: 1353–1382.

15.

Stefaniak

Dobrzynska

Drzymala-Czyz

, et al. Diet in the prevention of Alzheimer's disease: current knowledge and future research requirements. Nutrients 2022; 14: 4564.

16.

Zhong

Kuang

Danning

, et al. A genome-wide association study of bitter and sweet beverage consumption. Hum Mol Genet 2019; 28: 2449–2457.

17.

Coffee, Caffeine Genetics C; Cornelis

, et al. Genome-wide meta-analysis identifies six novel loci associated with habitual coffee consumption. Mol Psychiatry 2015; 20: 647–656.

18.

Meddens

SFW

de Vlaming

Bowers

, et al. Genomic analysis of diet composition finds novel loci and associations with health and lifestyle. Mol Psychiatry 2021; 26: 2056–2069.

19.

Liu

Jiang

Wedow

, et al. Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use. Nat Genet 2019; 51: 237–244.

20.

Cole

Florez

Hirschhorn

. Comprehensive genomic analysis of dietary habits in UK biobank identifies hundreds of genetic associations. Nat Commun 2020; 11: 1467.

21.

Birney

. Mendelian randomization. Cold Spring Harb Perspect Med 2022; 12: a041302.

22.

Meng

Chen

Zhu

, et al. Dietary factors and Alzheimer's disease risk: a Mendelian randomization study. Eur J Med Res 2024; 29: 261.

23.

Teng

Sun

Chen

, et al. Genetically determined dietary habits and risk of Alzheimer's disease: a Mendelian randomization study. Front Nutr 2024; 11: 1415555.

24.

Huang

Zhao

Shen

, et al. Causal relationship between dietary habits and Alzheimer's disease risk: a univariable and multivariable Mendelian randomization study. Food Funct 2025; 16: 1486–1494.

25.

Wang

Sun

Pan

, et al. Causal relationships between relative intake from the macronutrients and Alzheimer's disease: a two-sample Mendelian randomization study. J Alzheimers Dis 2022; 87: 665–673.

26.

Liao

Zhu

Xin

, et al. Fruit intake and Alzheimer's disease: results from Mendelian randomization. J Prev Alzheimers Dis 2024; 11: 445–452.

27.

Wightman

Jansen

Savage

, et al. A genome-wide association study with 1,126,563 individuals identifies new risk loci for Alzheimer's disease. Nat Genet 2021; 53: 1276–1282.

28.

Jansen

Savage

Watanabe

, et al. Genome-wide meta-analysis identifies new loci and functional pathways influencing Alzheimer's disease risk. Nat Genet 2019; 51: 404–413.

29.

Richmond

Davey Smith

. Mendelian randomization: concepts and scope. Cold Spring Harb Perspect Med 2022; 12: a040501.

30.

Burgess

Thompson

. Interpreting findings from Mendelian randomization using the MR-egger method. Eur J Epidemiol 2017; 32: 377–389.

31.

Verbanck

Chen

Neale

, et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018; 50: 693–698.

32.

Burgess

Foley

Allara

, et al. A robust and efficient method for Mendelian randomization with hundreds of genetic variants. Nat Commun 2020; 11: 376.

33.

Chatterjee

. A comprehensive evaluation of methods for Mendelian randomization using realistic simulations and an analysis of 38 biomarkers for risk of type 2 diabetes. Int J Epidemiol 2021; 50: 1335–1349.

34.

Shapland

Zhao

Bowden

. Profile-likelihood Bayesian model averaging for two-sample summary data Mendelian randomization in the presence of horizontal pleiotropy. Stat Med 2022; 41: 1100–1119.

35.

Gkatzionis

Burgess

Conti

, et al. Bayesian variable selection with a pleiotropic loss function in Mendelian randomization. Stat Med 2021; 40: 5025–5045.

36.

Seng

Liu

Wang

, et al. Instrumental variable model average with applications in Mendelian randomization. Stat Med 2023; 42: 3547–3567.

37.

Samuelsson

Stubbendorff

Marseglia

, et al. A comparative study of the EAT-lancet diet and the Mediterranean diet in relation to neuroimaging biomarkers and cognitive performance. Alzheimers Dement 2025; 21: e70191.

38.

Hoscheidt

Sanderlin

Baker

, et al. Mediterranean and Western diet effects on Alzheimer's disease biomarkers, cerebral perfusion, and cognition in mid-life: a randomized trial. Alzheimers Dement 2022; 18: 457–468.

39.

Gutierrez

Folch

Rojas

, et al. Effects of nutrition on cognitive function in adults with or without cognitive impairment: a systematic review of randomized controlled clinical trials. Nutrients 2021; 13: 3728.

40.

Abbasi

Khandae

Shahabi

, et al. Association between the DASH diet and Alzheimer's disease in a case-control study. Sci Rep 2025; 15: 23312.

41.

Liu

Morris

Dhana

, et al. Mediterranean-DASH Intervention for neurodegenerative delay (MIND) study: rationale, design and baseline characteristics of a randomized control trial of the MIND diet on cognitive decline. Contemp Clin Trials 2021; 102: 106270.

42.

Seago

Davy

, et al. Neuroprotective dietary patterns and longitudinal changes in cognitive function in older adults. J Acad Nutr Diet 2025; 125: 785–795 e789.

43.

Nishi

Babio

Gomez-Martinez

, et al. Mediterranean, DASH, and MIND dietary patterns and cognitive function: the 2-year longitudinal changes in an older Spanish cohort. Front Aging Neurosci 2021; 13: 782067.

44.

Rimm

Smith-Warner

, et al. Reproducibility and validity of dietary patterns assessed with a food-frequency questionnaire. Am J Clin Nutr 1999; 69: 243–249.

45.

Yeh

Blacker

Willett

. Dietary factors and cognitive function: with a focus on subjective cognitive decline. Curr Nutr Rep 2025; 14: 62.

46.

Yeh

Yuan

Ascherio

, et al. Long-term intake of total energy and fat in relation to subjective cognitive decline. Eur J Epidemiol 2022; 37: 133–146.

47.

Yeh

Yuan

Ascherio

, et al. Long-term dietary protein intake and subjective cognitive decline in US men and women. Am J Clin Nutr 2022; 115: 199–210.

48.

Yeh

Yuan

Ascherio

, et al. Long-term dietary flavonoid intake and subjective cognitive decline in US men and women. Neurology 2021; 97: e1041–e1056.

49.

Wernicke

Pohrt

Pletsch-Borba

, et al. Effect of unsaturated fat and protein intake on liver fat in people at risk of unhealthy aging: 1-year results of a randomized controlled trial. Am J Clin Nutr 2023; 117: 785–793.

50.

Gao

Yang

Yan

, et al. Protein intake from different sources and cognitive decline over 9 years in community-dwelling older adults. Front Public Health 2022; 10: 1016016.

51.

Perez

You

Kendrick

. Association of plant-based protein intake with cognitive function in adults with CKD. Kidney360 2023; 4: 1554–1561.

52.

Fernando

Rainey-Smith

Gardener

, et al. Associations of dietary protein and fiber intake with brain and blood amyloid-beta. J Alzheimers Dis 2018; 61: 1589–1598.

53.

Fujiwara

Tanaka

Kobayashi

, et al. Analysis of the association between protein intake and disability-adjusted life year rates for Alzheimer's disease in Japanese aged over 60. Nutrients 2024; 16: 1221.

54.

You

. Global patterns linking total meat supply to dementia incidence: a population-based ecological study. AIMS Neurosci 2025; 12: 203–221.

55.

Parrella

Maxim

Maialetti

, et al. Protein restriction cycles reduce IGF-1 and phosphorylated tau, and improve behavioral performance in an Alzheimer's disease mouse model. Aging Cell 2013; 12: 257–268.

56.

Siddik

MAB

Mullins

Kramer

, et al. Branched-chain amino acids are linked with Alzheimer's disease-related pathology and cognitive deficits. Cells 2022; 11: 3523.

57.

Tournissac

Vandal

Tremblay

, et al. Dietary intake of branched-chain amino acids in a mouse model of Alzheimer's disease: effects on survival, behavior, and neuropathology. Alzheimers Dement (N Y) 2018; 4: 677–687.

58.

Bertoni

Pini

Mazzocchi

, et al. The role of alpha-linolenic acid and other polyunsaturated fatty acids in mental health: a narrative review. Int J Mol Sci 2024; 25: 12479.

59.

Eskelinen

Ngandu

Helkala

, et al. Fat intake at midlife and cognitive impairment later in life: a population-based CAIDE study. Int J Geriatr Psychiatry 2008; 23: 741–747.

60.

Costanzo

Orellana

Nowson

, et al. Fat taste sensitivity is associated with short-term and habitual fat intake. Nutrients 2017; 9: 781.

61.

Currenti

Godos

Alanazi

, et al. Dietary fats and cognitive status in Italian middle-old adults. Nutrients 2023; 15: 1429.

62.

Fan

Borenstein

Wang

, et al. Associations of circulating saturated long-chain fatty acids with risk of mild cognitive impairment and Alzheimer's disease in the Alzheimer's disease neuroimaging initiative (ADNI) cohort. EBioMedicine 2023; 97: 104818.

63.

Ashton

Roberts

Wakefield

, et al. Medium chain triglycerides with a C8:c10 ratio of 30:70 enhances cognitive performance and mitigates the cognitive decline associated with prolonged exercise in young and healthy adults. Physiol Behav 2023; 269: 114284.

64.

Norgren

Sindi

Sandebring-Matton

, et al. The dietary carbohydrate/fat-ratio and cognitive performance: panel analyses in older adults at risk for dementia. Curr Dev Nutr 2023; 7: 100096.

65.

Morris

Vidoni

Honea

, et al. Impaired glycemia increases disease progression in mild cognitive impairment. Neurobiol Aging 2014; 35: 585–589.

66.

Degen

Toro

Schonknecht

, et al. Diabetes mellitus type II and cognitive capacity in healthy aging, mild cognitive impairment and Alzheimer's disease. Psychiatry Res 2016; 240: 42–46.

67.

Gardener

Rainey-Smith

Sohrabi

, et al. Increased carbohydrate intake is associated with poorer performance in verbal memory and attention in an APOE genotype-dependent manner. J Alzheimers Dis 2017; 58: 193–201.

68.

Schaefer

Kaiser

Eichner

, et al. Association of sugar intake from different sources with incident dementia in the prospective cohort of UK biobank participants. Nutr J 2023; 22: 42.

69.

Taylor

Sullivan

Swerdlow

, et al. A high-glycemic diet is associated with cerebral amyloid burden in cognitively normal older adults. Am J Clin Nutr 2017; 106: 1463–1470.

70.

Seetharaman

Andel

McEvoy

, et al. Blood glucose, diet-based glycemic load and cognitive aging among dementia-free older adults. J Gerontol A Biol Sci Med Sci 2015; 70: 471–479.

71.

Kean

Lamport

Dodd

, et al. Chronic consumption of flavanone-rich orange juice is associated with cognitive benefits: an 8-wk, randomized, double-blind, placebo-controlled trial in healthy older adults. Am J Clin Nutr 2015; 101: 506–514.

72.

Chen

Wang

, et al. Sugar-sweetened beverages consumption is associated with worse cognitive functions in older adults: from the national health and nutrition examination survey and food patterns equivalents database. Nutr Neurosci 2023; 26: 1011–1018.

73.

Zhang

Xiao

Cheng

, et al. Associations of sugar intake, high-sugar dietary pattern, and the risk of dementia: a prospective cohort study of 210,832 participants. BMC Med 2024; 22: 298.

74.

Liu

Volpe

Ross

, et al. Dietary sugar intake and risk of Alzheimer's disease in older women. Nutr Neurosci 2022; 25: 2302–2313.

75.

Liu

Xia

, et al. High-fiber diet mitigates maternal obesity-induced cognitive and social dysfunction in the offspring via gut-brain axis. Cell Metab 2021; 33: 923–938 e926.

76.

Gao

Song

, et al. Effects of oat fiber intervention on cognitive behavior in LDLR(-/-) mice modeling atherosclerosis by targeting the microbiome-gut-brain axis. J Agric Food Chem 2020; 68: 14480–14491.

77.

Wei

Zhu

, et al. Low expression of the intestinal metabolite butyric acid and the corresponding memory pattern regulate HDAC4 to promote apoptosis in rat hippocampal neurons. Ecotoxicol Environ Saf 2023; 253: 114660.

78.

Prokopidis

Giannos

Ispoglou

, et al. Dietary fiber intake is associated with cognitive function in older adults: data from the national health and nutrition examination survey. Am J Med 2022; 135: e257–e262.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.18 MB

0.11 MB